diff --git a/.github/workflows/windows-smoke.yml b/.github/workflows/windows-smoke.yml
index 522f1414..8a1f787a 100644
--- a/.github/workflows/windows-smoke.yml
+++ b/.github/workflows/windows-smoke.yml
@@ -32,8 +32,15 @@ jobs:
       - name: Install package
         run: |
           python -m pip install --upgrade pip setuptools wheel
-          pip install agentscope
-          pip install -e .
+          pip install -e ".[dev,core]"
 
       - name: Run version job
         run: reme start service.backend=cli job=version
+
+      - name: Run Windows path tests
+        run: |
+          python -m pytest `
+            tests/unit/test_auto_dream.py::test_scan_day_files_includes_nested_md_and_excludes_interests `
+            tests/unit/test_auto_dream.py::test_dream_extract_matches_posix_catalog_paths `
+            tests/unit/test_read_with_neighbors.py::test_read_with_neighbors_uses_posix_nested_path `
+            -v
diff --git a/docs/en/auto_dream.md b/docs/en/auto_dream.md
index d84a24ab..749ebe46 100644
--- a/docs/en/auto_dream.md
+++ b/docs/en/auto_dream.md
@@ -9,7 +9,7 @@ units into `digest/`, and generates the day's `interests.yaml` for proactive use
 </p>
 
 Its daily inputs usually come from [Auto Memory](./auto_memory.md) and [Auto Resource](./auto_resource.md). For the file
-semantics of `digest/`, `derived_from::`, and wikilinks, see [Memory as File](./memory_as_file.md). For the linking strategy
+semantics of `digest/`, Sources sections, and wikilinks, see [Memory as File](./memory_as_file.md). For the linking strategy
 used during Integrate, see [Auto Link](./auto_link.md). To read `interests.yaml`, use [Proactive](./proactive.md).
 
 ## Configuration
@@ -198,8 +198,8 @@ jobs:
 `auto_dream` consumes only daily inputs and does not rewrite daily bodies. Daily preserves facts and the original situation;
 digest is the abstracted long-term memory layer.
 
-`digest` is not a copy of the source text. Its body should preserve reusable abstractions, while details point back to sources
-through `derived_from:: [[daily/<date>/...]]`. Links follow the workspace-relative wikilink semantics described in
+`digest` is not a copy of the source text. Its body should preserve reusable abstractions, while a Sources section points
+back with entries such as `- [[daily/<date>/...]]`. Links follow the workspace-relative wikilink semantics described in
 [Memory as File](./memory_as_file.md).
 
 `auto_dream` does not invent an overview from nothing. Only content that actually appears in daily input and is extracted as
diff --git a/docs/en/auto_link.md b/docs/en/auto_link.md
index dde6afcb..0a97976a 100644
--- a/docs/en/auto_link.md
+++ b/docs/en/auto_link.md
@@ -32,7 +32,7 @@ link to multiple sources and multiple related digest nodes.
 |---|---|
 | The same memory already exists | Recall and update the existing node instead of creating a duplicate. |
 | New and existing material are related | Write workspace-relative wikilinks into the body. |
-| A digest node is disconnected from its sources | Point back to daily/resource source material with `derived_from:: [[...]]`. |
+| A digest node is disconnected from its sources | Add daily/resource links under a `## Sources` section. |
 | A node contains only isolated prose | Add links to related digest nodes on both CREATE and UPDATE. |
 
 ## Toolchain
@@ -77,20 +77,22 @@ Every unit must select one action:
 | Action | Linking semantics |
 |---|---|
 | `CREATE` | Write a new `digest/<bucket>/<slug>.md` and add source and related-node links to its body. |
-| `CORROBORATE` | The same abstraction appeared again; append a new `derived_from:: [[...]]` and strengthen the description when needed. |
+| `CORROBORATE` | The same abstraction appeared again; append its source link and strengthen the description when needed. |
 | `REFINE` | New material extends the existing node; insert the additional content in the appropriate section and preserve existing links. |
 | `CORRECT` | New material corrects the existing node; use source links to identify the basis for the correction. |
 
-An UPDATE should be additive whenever possible: do not delete existing wikilinks or `derived_from` entries. This prevents
+An UPDATE should be additive whenever possible: do not delete existing wikilinks or source entries. This prevents
 later graph indexing and retrieval from losing edges.
 
 ### 3. Write source edges
 
-Source edges use Markdown wikilinks:
+Source edges are ordinary wikilinks grouped under a Markdown heading:
 
 ```markdown
-derived_from:: [[daily/2026-06-20/session.md]]
-derived_from:: [[resource/2026-06-20/paper.md]]
+## Sources
+
+- [[daily/2026-06-20/session.md]]
+- [[resource/2026-06-20/paper.md]]
 ```
 
 These edges represent the evidence behind a digest node. Plain-text descriptions do not count as source edges because only
@@ -99,17 +101,14 @@ wikilinks can be parsed reliably by the file graph. For the complete parsing rul
 
 ### 4. Write relationships between digest nodes
 
-Relationships between digest nodes also use complete workspace-relative paths:
+Relationships between digest nodes use complete workspace-relative paths woven into natural prose:
 
 ```markdown
-relates_to:: [[digest/wiki/hybrid-search.md]]
-depends_on:: [[digest/procedure/rebuild-index.md]]
-blocks_on:: [[digest/personal/team-review-preference.md]]
+This design extends [[digest/wiki/hybrid-search.md]] and uses
+[[digest/procedure/rebuild-index.md]]. Follow
+[[digest/personal/team-review-preference.md]] during review.
 ```
 
-Predicates are open-ended. Common forms include `relates_to::`, `depends_on::`, and `blocks_on::`. The predicate sits outside
-the brackets, while the target path goes inside `[[...]]` and should include the `.md` suffix.
-
 ## Bucket Differences
 
 `auto_link` adjusts the shape of its output according to the unit bucket:
diff --git a/docs/en/memory_as_file.md b/docs/en/memory_as_file.md
index 6be4d56d..da41506c 100644
--- a/docs/en/memory_as_file.md
+++ b/docs/en/memory_as_file.md
@@ -23,7 +23,7 @@ ReMe represents memory as files not merely for convenient storage, but to give l
 |---|---|
 | Readable | Users can open the workspace directly and read daily notes, digest nodes, and source material like ordinary notes. |
 | Editable | Users and agents can correct, extend, move, or delete memory with file operations, without a specialized database client. |
-| Traceable | Long-term conclusions in digest can point back to daily, resource, or session sources through `derived_from:: [[...]]`. |
+| Traceable | Long-term conclusions in digest can point back to daily, resource, or session files from a Sources section. |
 | Portable | The workspace is an ordinary directory. Markdown, JSONL, YAML, and resource files can be backed up, synchronized, versioned, or moved to other tools. |
 | Indexable | Although the files are plain text, ReMe parses frontmatter, chunks, and wikilinks to build a retrieval index and file graph. |
 | Collaborative | Humans judge and correct; agents organize, link, and retrieve. Both operate on the same files. |
@@ -131,9 +131,7 @@ tags: [new energy, solar]
 # Conclusions
 
 The solar supply chain consists of [[digest/wiki/polysilicon.md]], wafers, cells, and modules.
-
-upstream:: [[digest/wiki/polysilicon.md]]
-[company:: [[digest/wiki/longi.md|LONGi]]]
+One major producer is [[digest/wiki/longi.md|LONGi]].
 ```
 
 ### Frontmatter
@@ -163,15 +161,18 @@ confidence: observed
 
 The user repeatedly asks documentation to explain motivation, boundaries, and examples while avoiding marketing language.
 
-derived_from:: [[daily/2026-06-20/session-a.md]]
-related:: [[digest/procedure/technical-documentation.md]]
+Apply this preference when following [[digest/procedure/technical-documentation.md]].
+
+## Sources
+
+- [[daily/2026-06-20/session-a.md]]
 ```
 
 This has three benefits:
 
 1. `name` and `description` serve as lightweight summaries in lists, recall results, and agent decisions.
 2. The body can carry fuller facts, conditions, counterexamples, and sources.
-3. Typed wikilinks such as `derived_from::` and `related::` can be parsed by the graph and maintained when files move.
+3. Ordinary wikilinks can be parsed by the graph and maintained when files move.
 
 Frontmatter is best for stable, short, structured fields; the body is best for explanations meant for people. Do not put long
 body text into YAML fields.
@@ -181,10 +182,10 @@ body text into YAML fields.
 Wikilinks express relationships between files with `[[...]]`:
 
 ```text
-[[digest/wiki/solar.md]]
-[[digest/wiki/solar.md#supply-chain]]
-[[digest/wiki/solar.md|solar]]
-![[resource/2026-06-01/report.md]]
+[[daily/2026-06-20/session.md]]
+[[notes/example.md#L9]]
+[[notes/example.md#L9-L10]]
+[[notes/example.md#L9-L10,L15-L20]]
 ```
 
 ReMe wikilinks use **literal path semantics**:
@@ -196,54 +197,59 @@ ReMe wikilinks use **literal path semantics**:
 ReMe does not append `.md` automatically, search by filename, or automatically resolve folder notes. Use complete
 workspace-relative paths with their extensions.
 
+Ordinary Markdown links such as `[label](../wiki/example.md)` do not create `FileLink` edges and are not rewritten by move or
+retarget operations.
+
+Anchors such as `#L9`, `#L9-L10`, and `#L9-L10,L15-L20` remain ordinary `target_anchor` strings in the graph. The graph
+parser does not validate line-anchor syntax, so values such as `#L0`, `#L10-L9`, and `#L9,` are also stored. The `read` job
+does not interpret an anchor appended to `path`; use the separate 1-based, inclusive `start_line` and `end_line` arguments to
+read a range, for example `read(path="digest/wiki/solar.md", start_line=9, end_line=10)`.
+
 Wikilinks support these behaviors:
 
 ```text
 body link          -> create a FileLink
-predicate:: link   -> create a FileLink with a relationship name
 move a file        -> rewrite [[old path]] in inbound edges by default
 delete a file      -> return remaining inbound edges so references can be cleaned up
 search match       -> expand inbound and outbound links to provide context
 ```
 
-Supported relationship forms:
-
-```markdown
-industry:: [[digest/wiki/new-energy.md]]
-[competitor:: [[digest/wiki/byd.md]]]
-```
-
 Parsed result:
 
 ```text
 FileLink
   source_path = current file
-  target_path = digest/wiki/new-energy.md
-  predicate   = industry
+  target_path = notes/example.md
+  target_anchor = L9-L10,L15-L20
 ```
 
+Older documents containing wrappers such as `related:: [[path]]`,
+`- related:: [[path]]`, or `[related:: [[path]]]` remain readable. ReMe
+ignores the surrounding text and indexes the inner `[[path]]` as an ordinary
+link. After upgrading from a version that stored typed links, run `reme reindex`
+once to rebuild the derived graph without the removed relationship field.
+
 ### Sources and Relationships
 
 The two most important link types in ReMe are source links and conceptual relationship links.
 
-A source link explains where a long-term memory came from:
+A Sources section records where a long-term memory came from:
 
 ```markdown
-derived_from:: [[daily/2026-06-20/session-a.md]]
-derived_from:: [[resource/2026-06-20/report.pdf]]
+## Sources
+
+- [[daily/2026-06-20/session-a.md]]
+- [[resource/2026-06-20/report.pdf]]
 ```
 
-A conceptual relationship link explains which other long-term memories relate to the node:
+A conceptual relationship link explains which other long-term memories relate to the node. Weave it into natural prose:
 
 ```markdown
-related:: [[digest/wiki/solar-supply-chain.md]]
-depends_on:: [[digest/procedure/research-report-analysis.md]]
-contrasts_with:: [[digest/wiki/central-inverter.md]]
+This analysis extends [[digest/wiki/solar-supply-chain.md]], follows
+[[digest/procedure/research-report-analysis.md]], and contrasts with
+[[digest/wiki/central-inverter.md]].
 ```
 
-Ordinary body wikilinks also create graph edges, but when the relationship itself has semantic value, prefer
-`predicate:: [[path]]`. This makes the meaning of links clearer to search, graph traversal, and later agent integration.
-
 ## Human and Agent Editing
 
 Because memory is stored as files, users can edit the workspace directly, while agents can read and write the same files
@@ -257,8 +263,8 @@ through ReMe's file tools. Both follow the same conventions:
 | Delete a file | Check inbound links first. ReMe's delete tool returns source files that still point to the target, making dangling references easier to clean up. |
 | Edit metadata | Use frontmatter for short fields. When the body changes substantially, update `description` as well. |
 
-A practical rule is: **an agent may rewrite the wording, but it must not lose evidence edges**. In particular,
-`derived_from:: [[...]]` and existing digest-to-digest wikilinks are the basis for traceable and extensible long-term memory.
+A practical rule is: **an agent may rewrite the wording, but it must not lose evidence edges**. In particular, Sources entries
+and existing digest-to-digest wikilinks are the basis for traceable and extensible long-term memory.
 
 ## Path Semantics
 
diff --git a/docs/en/memory_search.md b/docs/en/memory_search.md
index 9ee3605a..5f2c3784 100644
--- a/docs/en/memory_search.md
+++ b/docs/en/memory_search.md
@@ -66,7 +66,7 @@ stable batch of changes.
 4. For a deleted file, remove its records from `file_store`, `keyword_index`, and `file_graph`.
 5. When changes exist, dump state to `metadata/` so it can be restored on the next startup.
 
-The Markdown chunker parses YAML frontmatter, heading structure, and `[[...]]` into `FileNode`, `FileChunk`, and `FileLink`
+The Markdown chunker parses YAML frontmatter, heading structure, and wikilinks into `FileNode`, `FileChunk`, and `FileLink`
 objects. For detailed chunking rules, see [Memory as File](./memory_as_file.md#memory-chunking).
 
 ### Index Optimization
@@ -182,7 +182,8 @@ and then adds the new chunk text. Deletion is lazy; the index can later be compa
 "Progressive" in Memory Search does not mean putting the entire repository into one result. Retrieval expands in three layers:
 
 1. Chunk recall: return only the `limit` most relevant text fragments.
-2. File location: each result includes `path:start_line-end_line`, allowing the caller to read the source precisely with `read`.
+2. File location: each result includes `path:start_line-end_line`. Pass the path and line bounds separately as `path`,
+   `start_line`, and `end_line` when calling `read`; the range is not part of the `path` value.
 3. Link neighbors: call `expand_links()` for each matched file and expand at most `max_links_per_direction` outlinks and
    inlinks.
 
@@ -194,7 +195,7 @@ matched chunk
   -> file_store.get_outlinks(path)
   -> file_store.get_inlinks(path)
   -> file_store.get_nodes(neighbor_paths)
-  -> render neighbor path, name, description, predicate, and anchor
+  -> render neighbor path, name, description, and anchor
 ```
 
 This keeps search results short while still showing which long-term nodes, resources, or other daily notes a memory connects
@@ -216,10 +217,8 @@ Typical text structure:
 ...matched memory fragment...
   outlinks (2):
     -> digest/indexing.md  name="Indexing"  description="..."
-      via predicate=related
   inlinks (1):
     <- daily/2026-06-19.md  name="..."
-      via plain
 ```
 
 `counts` reports how many vector and keyword candidates were recalled and how many results were ultimately returned. With
diff --git a/docs/en/reme_scene.md b/docs/en/reme_scene.md
index 3ba652bd..d7ce86d5 100644
--- a/docs/en/reme_scene.md
+++ b/docs/en/reme_scene.md
@@ -110,17 +110,19 @@ name: Cobalt
 description: A key raw material for lithium-battery cathodes, with production concentrated in the DRC
 ---
 
-downstream_product:: [[digest/wiki/ternary-cathodes.md]]
-producer:: [[digest/wiki/glencore.md]]
-source_event:: [[daily/2026-05-18/2026-05-18-close.md]]
-
 # Cobalt
 
+Used by [[digest/wiki/ternary-cathodes.md]]; a major producer is [[digest/wiki/glencore.md]].
+
 ## Supply
 Glencore's third-quarter cobalt output fell 18% year over year. Continue monitoring how tighter supply affects prices.
 
 ## Policy risk
 Changes to mining-rights policy in the DRC may affect KFM mine operations and should be tracked together with CMOC.
+
+## Sources
+
+- [[daily/2026-05-18/2026-05-18-close.md]]
 ```
 
 Note that wikilinks use literal path semantics. Prefer complete workspace-relative paths with the `.md` extension. ReMe
@@ -185,10 +187,10 @@ The result shape is:
 Glencore's third-quarter cobalt output fell 18% year over year...
 
   outlinks:
-    -> digest/wiki/ternary-cathodes.md name="Ternary Cathodes" via predicate=downstream_product
-    -> digest/wiki/glencore.md name="Glencore" via predicate=producer
+    -> digest/wiki/ternary-cathodes.md name="Ternary Cathodes"
+    -> digest/wiki/glencore.md name="Glencore"
   inlinks:
-    <- digest/wiki/ternary-cathodes.md name="Ternary Cathodes" via predicate=upstream_material
+    <- digest/wiki/ternary-cathodes.md name="Ternary Cathodes"
 
 ========== digest/wiki/ternary-cathodes.md:5-18 [score=0.0139 keyword=3.2017] ==========
 ...
@@ -301,11 +303,10 @@ name: TypeScript project build OOM diagnostic path
 description: When a build stalls and memory grows, check the type-checking process first
 ---
 
-source_event:: [[daily/2026-03-10/build-oom-2026-03-10.md]]
-related_preference:: [[digest/personal/code-style.md]]
-
 # TypeScript Project Build OOM Diagnostic Path
 
+Apply [[digest/personal/code-style.md]] while following this runbook.
+
 ## Symptoms
 The build stalls near the end. CPU usage is low, but memory keeps growing.
 
@@ -317,6 +318,10 @@ The build stalls near the end. CPU usage is low, but memory keeps growing.
 ## Known ineffective paths
 - Deleting `.cache` alone did not resolve the issue on 2026-03-10.
 - Upgrading the terser plugin did not resolve the issue on 2026-03-10.
+
+## Sources
+
+- [[daily/2026-03-10/build-oom-2026-03-10.md]]
 ```
 
 Example `digest/personal/code-style.md`:
@@ -411,13 +416,14 @@ name: Alice
 description: A friend of the user who often recommends reading material
 ---
 
-recommended_book:: [[digest/wiki/deep-work.md]]
-source_event:: [[daily/2026-04-20/lunch-with-alice.md]]
-
 # Alice
 
 ## Reading recommendations
-At lunch on 2026-04-20, Alice recommended a book about attention and deep work.
+At lunch on 2026-04-20, Alice recommended [[digest/wiki/deep-work.md]], a book about attention and deep work.
+
+## Sources
+
+- [[daily/2026-04-20/lunch-with-alice.md]]
 ```
 
 ### An associative recall
@@ -439,7 +445,7 @@ Matches:
 ```text
 digest/personal/alice.md
   outlinks:
-    -> digest/wiki/deep-work.md via predicate=recommended_book
+    -> digest/wiki/deep-work.md
 daily/2026-04-20/lunch-with-alice.md
 ```
 
diff --git a/docs/zh/auto_dream.md b/docs/zh/auto_dream.md
index 315c92b9..333f58ed 100644
--- a/docs/zh/auto_dream.md
+++ b/docs/zh/auto_dream.md
@@ -7,7 +7,7 @@
   <img src="../figure/auto-dream-and-proactive.svg" alt="ReMe Auto Dream and Proactive 从 daily 到 digest 再到 proactive 的流程" width="92%">
 </p>
 
-它消费的 daily 输入通常来自 [Auto Memory](./auto_memory.md) 和 [Auto Resource](./auto_resource.md)。`digest/`、`derived_from::`
+它消费的 daily 输入通常来自 [Auto Memory](./auto_memory.md) 和 [Auto Resource](./auto_resource.md)。`digest/`、Sources 章节
 和 wikilink 的文件语义见 [Memory as File](./memory_as_file.md)；Integrate 阶段的链接策略详见 [Auto Link](./auto_link.md)。
 `interests.yaml` 的读取接口见 [Proactive](./proactive.md)。
 
@@ -189,7 +189,7 @@ jobs:
 
 `auto_dream` 只消费 daily 输入，不改写 daily 正文。daily 是事实和现场记录，digest 才是抽象后的长期记忆层。
 
-`digest` 不是原文复制。正文应保留可复用抽象，细节通过 `derived_from:: [[daily/<date>/...]]` 指回来源。链接写法遵循
+`digest` 不是原文复制。正文应保留可复用抽象，细节通过 Sources 章节中的 `- [[daily/<date>/...]]` 指回来源。链接写法遵循
 [Memory as File](./memory_as_file.md) 中的 workspace-relative wikilink 语义。
 
 `auto_dream` 不凭空生成总览。只有 daily 输入中确实出现、并被抽取为 unit 或 topic 的内容，才会进入 digest 或
diff --git a/docs/zh/auto_link.md b/docs/zh/auto_link.md
index ae50bb0b..c34c8423 100644
--- a/docs/zh/auto_link.md
+++ b/docs/zh/auto_link.md
@@ -30,7 +30,7 @@ Integrate 阶段对每个 unit 独立运行。一个 unit 只落到一个目标
 |--------------|----------------------------------------------------|
 | 已有相同记忆       | 召回后更新旧节点，而不是重复创建。                                  |
 | 新旧材料有关联      | 在正文中写入 workspace-relative wikilink。                    |
-| digest 与来源断开 | 用 `derived_from:: [[...]]` 指回 daily/resource 原始材料。 |
+| digest 与来源断开 | 在 `## Sources` 章节加入指向 daily/resource 的链接。 |
 | 节点只有孤立正文     | 在 CREATE 和 UPDATE 时都补充相关 digest 节点链接。              |
 
 ## 工具链
@@ -73,19 +73,21 @@ Agent 先用 unit 的触发条件、动词、名词、同义词和可能的 fail
 | 动作            | 链接含义                                                |
 |---------------|-----------------------------------------------------|
 | `CREATE`      | 写入新的 `digest/<bucket>/<slug>.md`，并在新正文里加入来源和相关节点链接。 |
-| `CORROBORATE` | 同一抽象再次出现，追加新的 `derived_from:: [[...]]`，必要时强化描述。     |
+| `CORROBORATE` | 同一抽象再次出现，追加来源链接，必要时强化描述。     |
 | `REFINE`      | 新材料扩展了旧节点，把补充内容插入合适段落，并保留旧链接。                       |
 | `CORRECT`     | 新材料修正旧节点，用来源链接标出修正依据。                               |
 
-UPDATE 必须尽量只增不删：不要删除已有 wikilink 或 `derived_from`。这是为了让后续图谱索引和检索不会丢边。
+UPDATE 必须尽量只增不删：不要删除已有 wikilink 或来源条目。这是为了让后续图谱索引和检索不会丢边。
 
 ### 3. 写来源边
 
-来源边使用 markdown wikilink：
+来源边是归档在 Markdown 固定章节下的普通 Wikilink：
 
 ```markdown
-derived_from:: [[daily/2026-06-20/session.md]]
-derived_from:: [[resource/2026-06-20/paper.md]]
+## Sources
+
+- [[daily/2026-06-20/session.md]]
+- [[resource/2026-06-20/paper.md]]
 ```
 
 这些边表示 digest 节点的证据来源。纯文本描述不算来源边，因为只有 wikilink 能被 file graph 稳定解析。更完整的 wikilink
@@ -93,17 +95,14 @@ derived_from:: [[resource/2026-06-20/paper.md]]
 
 ### 4. 写 digest 关联边
 
-digest 之间的关联也使用完整 workspace-relative 路径：
+digest 之间的关联使用完整 workspace-relative 路径，并自然织入正文：
 
 ```markdown
-relates_to:: [[digest/wiki/hybrid-search.md]]
-depends_on:: [[digest/procedure/rebuild-index.md]]
-blocks_on:: [[digest/personal/team-review-preference.md]]
+这个设计扩展了 [[digest/wiki/hybrid-search.md]]，并使用
+[[digest/procedure/rebuild-index.md]]。评审时遵循
+[[digest/personal/team-review-preference.md]]。
 ```
 
-谓词是开放的，常见写法包括 `relates_to::`、`depends_on::`、`blocks_on::`。谓词在括号外，目标路径在 `[[...]]` 内，并且应包含
-`.md` 后缀。
-
 ## Bucket 差异
 
 `auto_link` 的规则会随 unit bucket 调整写入形态：
diff --git a/docs/zh/memory_as_file.md b/docs/zh/memory_as_file.md
index 1be2de34..c38fb0db 100644
--- a/docs/zh/memory_as_file.md
+++ b/docs/zh/memory_as_file.md
@@ -22,7 +22,7 @@ ReMe 把记忆设计成文件，不只是为了“方便存储”，而是为了
 |----------|----------------------------------------------------------------------|
 | 可读       | 用户可以直接打开 workspace，像读普通笔记一样读 daily、digest 和原始材料。                         |
 | 可编辑      | 用户和 Agent 都能用文件操作修正、补充、移动或删除记忆，不必依赖专用数据库客户端。                    |
-| 可追溯      | digest 中的长期结论可以通过 `derived_from:: [[...]]` 回到 daily、resource 或 session 原文。 |
+| 可追溯      | digest 中的长期结论可以通过 Sources 章节回到 daily、resource 或 session 原文。 |
 | 可迁移      | workspace 是普通目录，Markdown、JSONL、YAML 和资源文件可以被备份、同步、版本管理或迁移到其他工具。          |
 | 可索引      | 文件虽然是普通文本，但 ReMe 会解析 frontmatter、chunk、wikilink，构建检索索引和文件图谱。         |
 | 可协作      | 人负责判断和修正，Agent 负责整理、链接和检索；二者看到和操作的是同一套文件。                       |
@@ -124,9 +124,7 @@ tags: [新能源, 光伏]
 # 结论
 
 光伏产业链可以拆成 [[digest/wiki/硅料.md]]、硅片、电池片和组件。
-
-upstream:: [[digest/wiki/硅料.md]]
-[company:: [[digest/wiki/隆基绿能.md|隆基]]]
+主要生产商包括 [[digest/wiki/隆基绿能.md|隆基]]。
 ```
 
 ### Frontmatter
@@ -156,15 +154,18 @@ confidence: observed
 
 用户多次要求文档补充动机、边界和例子，但避免营销式表述。
 
-derived_from:: [[daily/2026-06-20/session-a.md]]
-related:: [[digest/procedure/技术文档写作.md]]
+执行 [[digest/procedure/技术文档写作.md]] 时应用这个偏好。
+
+## Sources
+
+- [[daily/2026-06-20/session-a.md]]
 ```
 
 这样做有三个好处：
 
 1. `name` 和 `description` 可以在列表、召回结果和 Agent 判断中作为轻量摘要。
 2. 正文可以承载更完整的事实、条件、反例和来源。
-3. `derived_from::`、`related::` 这类 typed wikilink 可以被图谱解析，后续移动文件时也能被维护。
+3. 普通 Wikilink 可以被图谱解析，后续移动文件时也能被维护。
 
 Frontmatter 适合放稳定、短小、结构化的字段；正文适合放需要人读的解释。不要把大段正文塞进 YAML 字段。
 
@@ -173,10 +174,10 @@ Frontmatter 适合放稳定、短小、结构化的字段；正文适合放需
 Wikilink 用 `[[...]]` 表达文件之间的关系：
 
 ```text
-[[digest/wiki/光伏.md]]
-[[digest/wiki/光伏.md#产业链]]
-[[digest/wiki/光伏.md|光伏]]
-![[resource/2026-06-01/report.md]]
+[[daily/2026-06-20/session.md]]
+[[notes/example.md#L9]]
+[[notes/example.md#L9-L10]]
+[[notes/example.md#L9-L10,L15-L20]]
 ```
 
 ReMe 的 wikilink 是**字面路径语义**：
@@ -187,54 +188,57 @@ ReMe 的 wikilink 是**字面路径语义**：
 
 它不会自动补 `.md`，不会按文件名搜索，也不会自动解析 folder note。推荐写完整的 workspace 相对路径，并带上扩展名。
 
+`[label](../wiki/example.md)` 这类普通 Markdown 链接不会建立 `FileLink`，move 或 retarget 操作也不会改写它们。
+
+`#L9`、`#L9-L10` 和 `#L9-L10,L15-L20` 这类锚点会作为普通 `target_anchor` 字符串保存在图谱中。图谱解析器
+不会校验行号锚点，因此 `#L0`、`#L10-L9`、`#L9,` 也会被保存。`read` 不会解析追加在 `path` 后的锚点；读取指定
+范围时需要分别传入从 1 开始、首尾均包含的 `start_line` 和 `end_line`，例如
+`read(path="digest/wiki/光伏.md", start_line=9, end_line=10)`。
+
 Wikilink 的作用：
 
 ```text
 正文链接       -> 建立 FileLink
-predicate:: 链接 -> 建立带关系名的 FileLink
 move 文件      -> 默认改写入边中的 [[旧路径]]
 delete 文件    -> 返回仍存在的入边，提示清理引用
 search 命中    -> 可展开出入链，帮助理解上下文
 ```
 
-支持的关系写法：
-
-```markdown
-industry:: [[digest/wiki/新能源.md]]
-[competitor:: [[digest/wiki/比亚迪.md]]]
-```
-
 解析结果：
 
 ```text
 FileLink
   source_path = 当前文件
-  target_path = digest/wiki/新能源.md
-  predicate   = industry
+  target_path = notes/example.md
+  target_anchor = L9-L10,L15-L20
 ```
 
+旧文档中的 `related:: [[path]]`、`- related:: [[path]]` 或
+`[related:: [[path]]]` 仍然可以读取。ReMe 会忽略外围文本，把内部 `[[path]]`
+作为普通链接建立索引。从曾存储 typed link 的版本升级后，应执行一次 `reme reindex`，
+用源文件重建不含旧关系字段的派生图索引。
+
 ### 来源和关系
 
 ReMe 里最重要的两类链接是来源链接和概念关系链接。
 
-来源链接说明“这条长期记忆从哪里来”：
+Sources 章节说明“这条长期记忆从哪里来”：
 
 ```markdown
-derived_from:: [[daily/2026-06-20/session-a.md]]
-derived_from:: [[resource/2026-06-20/report.pdf]]
+## Sources
+
+- [[daily/2026-06-20/session-a.md]]
+- [[resource/2026-06-20/report.pdf]]
 ```
 
-概念关系链接说明“这个节点和哪些长期记忆有关”：
+概念关系链接说明“这个节点和哪些长期记忆有关”，并自然织入正文：
 
 ```markdown
-related:: [[digest/wiki/光伏产业链.md]]
-depends_on:: [[digest/procedure/调研报告拆解流程.md]]
-contrasts_with:: [[digest/wiki/集中式逆变器.md]]
+这份分析扩展了 [[digest/wiki/光伏产业链.md]]，遵循
+[[digest/procedure/调研报告拆解流程.md]]，并与
+[[digest/wiki/集中式逆变器.md]] 对比。
 ```
 
-普通正文 wikilink 也会建立图边，但当关系本身有语义价值时，推荐使用 `predicate:: [[path]]`。这能让搜索、图遍历和后续 Agent
-整合更容易理解链接含义。
-
 ## 人工编辑和 Agent 编辑
 
 因为记忆就是文件，用户可以直接在编辑器里改 workspace；Agent 也可以通过 ReMe 的文件工具读写同一批文件。两者遵守同一套约定：
@@ -247,8 +251,8 @@ contrasts_with:: [[digest/wiki/集中式逆变器.md]]
 | 删除文件   | 删除前检查入链；ReMe 的 delete 会返回仍然指向目标的来源文件，方便清理悬空引用。                 |
 | 修改元数据  | 用 frontmatter 表达短字段；正文发生实质变化时同步更新 `description`。                  |
 
-一个实用规则是：**可以让 Agent 重写表达，但不要让它丢掉证据边**。尤其是 digest 节点中的 `derived_from:: [[...]]` 和已有
-digest-to-digest wikilink，是长期记忆可追溯和可扩展的基础。
+一个实用规则是：**可以让 Agent 重写表达，但不要让它丢掉证据边**。尤其是 digest 节点中的 Sources 条目和已有
+digest-to-digest Wikilink，是长期记忆可追溯和可扩展的基础。
 
 ## 路径语义
 
diff --git a/docs/zh/memory_search.md b/docs/zh/memory_search.md
index 483bf0d4..e97c42d4 100644
--- a/docs/zh/memory_search.md
+++ b/docs/zh/memory_search.md
@@ -62,7 +62,7 @@ index_update_loop:
 4. 对删除的文件，从 `file_store`、`keyword_index` 和 `file_graph` 清掉对应记录。
 5. 有变化时 dump 到 `metadata/`，让下次启动可以恢复。
 
-Markdown chunker 会解析 YAML frontmatter、标题结构和 `[[...]]`，产出 `FileNode`、`FileChunk` 和 `FileLink`。更细的分块规则见
+Markdown chunker 会解析 YAML frontmatter、标题结构和 wikilink，产出 `FileNode`、`FileChunk` 和 `FileLink`。更细的分块规则见
 [Memory as File](./memory_as_file.md#memory-chunking)。
 
 ### 索引优化
@@ -172,7 +172,8 @@ delete，后续可通过 optimize 压缩索引。
 Memory Search 的“渐进式”不是一次把全库内容塞进结果，而是分三层展开：
 
 1. 第一层是 chunk 召回：只返回最相关的 `limit` 个文本片段。
-2. 第二层是文件定位：每个结果带 `path:start_line-end_line`，可以继续用 `read` 精读原文件。
+2. 第二层是文件定位：每个结果带 `path:start_line-end_line`。调用 `read` 时需要把它们分别作为 `path`、
+   `start_line` 和 `end_line` 传入，行号范围不是 `path` 的一部分。
 3. 第三层是链接邻居：对命中文件调用 `expand_links()`，展开最多 `max_links_per_direction` 个 outlinks 和 inlinks。
 
 展开的数据来自 `file_graph`，不是重新扫文件：
@@ -183,7 +184,7 @@ Memory Search 的“渐进式”不是一次把全库内容塞进结果，而是
   -> file_store.get_outlinks(path)
   -> file_store.get_inlinks(path)
   -> file_store.get_nodes(neighbor_paths)
-  -> 渲染邻居的 path、name、description、predicate、anchor
+  -> 渲染邻居的 path、name、description、anchor
 ```
 
 这让搜索结果既保持短，又能看到“这条记忆连接到哪些长期节点、资源或其他 daily note”。如果某条结果值得继续追，可以用
@@ -203,10 +204,8 @@ Memory Search 的“渐进式”不是一次把全库内容塞进结果，而是
 ...命中的记忆片段...
   outlinks (2):
     -> digest/indexing.md  name="Indexing"  description="..."
-      via predicate=related
   inlinks (1):
     <- daily/2026-06-19.md  name="..."
-      via plain
 ```
 
 `counts` 会告诉你本次向量、关键词各召回了多少候选，以及最终返回多少条。默认 embedding 关闭时，`vector` 通常是 `0`，`hybrid` 是
diff --git a/docs/zh/reme_scene.md b/docs/zh/reme_scene.md
index a658f586..a2f203f5 100644
--- a/docs/zh/reme_scene.md
+++ b/docs/zh/reme_scene.md
@@ -107,15 +107,17 @@ name: 钴
 description: 锂电正极材料关键原料，主产区集中于刚果(金)
 ---
 
-downstream_product:: [[digest/wiki/三元正极.md]]
-producer:: [[digest/wiki/嘉能可.md]]
-source_event:: [[daily/2026-05-18/2026-05-18-close.md]]
-
 # 钴
 
+用于 [[digest/wiki/三元正极.md]]；主要生产商包括 [[digest/wiki/嘉能可.md]]。
+
 ## 供给端
 嘉能可三季度钴产量同比下滑 18%，需要继续跟踪供给收缩对价格的影响。
 
+## Sources
+
+- [[daily/2026-05-18/2026-05-18-close.md]]
+
 ## 政策风险
 刚果(金)矿权政策变化可能影响 KFM 矿运营，需联动跟踪洛阳钼业。
 ```
@@ -179,10 +181,10 @@ reme search query="锂电 上下游 三元 正极 钴 宁德" limit=5
 嘉能可三季度钴产量同比下滑 18%...
 
   outlinks:
-    -> digest/wiki/三元正极.md name="三元正极" via predicate=downstream_product
-    -> digest/wiki/嘉能可.md name="嘉能可" via predicate=producer
+    -> digest/wiki/三元正极.md name="三元正极"
+    -> digest/wiki/嘉能可.md name="嘉能可"
   inlinks:
-    <- digest/wiki/三元正极.md name="三元正极" via predicate=upstream_material
+    <- digest/wiki/三元正极.md name="三元正极"
 
 ========== digest/wiki/三元正极.md:5-18 [score=0.0139 keyword=3.2017] ==========
 ...
@@ -291,11 +293,10 @@ name: TypeScript 项目构建 OOM 排查路径
 description: build 卡住且内存上涨时，优先检查类型检查进程内存
 ---
 
-source_event:: [[daily/2026-03-10/build-oom-2026-03-10.md]]
-related_preference:: [[digest/personal/code-style.md]]
-
 # TypeScript 项目构建 OOM 排查路径
 
+执行这个 runbook 时遵循 [[digest/personal/code-style.md]]。
+
 ## 症状
 构建卡在后段，CPU 不高但内存持续上涨。
 
@@ -307,6 +308,10 @@ related_preference:: [[digest/personal/code-style.md]]
 ## 已知无效路径
 - 单纯删除 `.cache` 未解决 2026-03-10 的问题。
 - 升级 terser 插件未解决 2026-03-10 的问题。
+
+## Sources
+
+- [[daily/2026-03-10/build-oom-2026-03-10.md]]
 ```
 
 示例 `digest/personal/code-style.md`：
@@ -398,13 +403,14 @@ name: Alice
 description: 用户朋友，常推荐阅读材料
 ---
 
-recommended_book:: [[digest/wiki/deep-work.md]]
-source_event:: [[daily/2026-04-20/lunch-with-alice.md]]
-
 # Alice
 
 ## 阅读推荐
-2026-04-20 午餐时推荐过一本关于注意力和深度工作的书。
+2026-04-20 午餐时推荐过 [[digest/wiki/deep-work.md]]，这是一本关于注意力和深度工作的书。
+
+## Sources
+
+- [[daily/2026-04-20/lunch-with-alice.md]]
 ```
 
 ### 一次联想式回忆
@@ -426,7 +432,7 @@ reme search query="Alice 推荐 注意力 书 深度" limit=5
 ```text
 digest/personal/alice.md
   outlinks:
-    -> digest/wiki/deep-work.md via predicate=recommended_book
+    -> digest/wiki/deep-work.md
 daily/2026-04-20/lunch-with-alice.md
 ```
 
diff --git a/reme/components/base_component.py b/reme/components/base_component.py
index 2b55329a..dcb67b19 100644
--- a/reme/components/base_component.py
+++ b/reme/components/base_component.py
@@ -42,12 +42,12 @@ class ComponentMixin:
         return Path(self.app_context.app_config.workspace_dir).absolute()
 
     def to_workspace_relative(self, path: str | Path) -> str:
-        """Convert `path` to a workspace-relative string; return absolute path when outside."""
+        """Convert ``path`` to a POSIX workspace path; keep outside paths absolute."""
         abs_path = Path(path).absolute()
         try:
-            return str(abs_path.relative_to(self.workspace_path))
+            return abs_path.relative_to(self.workspace_path).as_posix()
         except ValueError:
-            return str(abs_path)
+            return abs_path.as_posix()
 
 
 class Dependency:
diff --git a/reme/components/file_chunker/default_file_chunker.py b/reme/components/file_chunker/default_file_chunker.py
index a2c57f3c..23457f5b 100644
--- a/reme/components/file_chunker/default_file_chunker.py
+++ b/reme/components/file_chunker/default_file_chunker.py
@@ -73,7 +73,7 @@ class DefaultFileChunker(BaseFileChunker):
         )
 
     def _link_byte_spans(self, content: str) -> list[tuple[int, int]]:
-        """Return [start, end) byte spans of every wikilink in content."""
+        """Return byte spans of wikilinks."""
         spans: list[tuple[int, int]] = []
         last_char, last_byte = 0, 0
         for wm in WikilinkHandler.iter_matches(content):
diff --git a/reme/components/file_chunker/markdown_file_chunker.py b/reme/components/file_chunker/markdown_file_chunker.py
index 1365020c..7da94ae6 100644
--- a/reme/components/file_chunker/markdown_file_chunker.py
+++ b/reme/components/file_chunker/markdown_file_chunker.py
@@ -12,8 +12,7 @@ and merge adjacent small subtrees at their parent. Leaf blocks (table / code /
 list / paragraph) split on internal boundaries and each piece is annotated
 ``[Part X/N]``. Wikilink extraction is
 delegated to :class:`reme.utils.wikilink_handler.WikilinkHandler` —
-the single source of truth for ``[[...]]`` syntax (including
-Dataview-style typed predicates).
+the single source of truth for ``[[...]]`` syntax.
 """
 
 import re
diff --git a/reme/components/file_graph/neo4j_file_graph.py b/reme/components/file_graph/neo4j_file_graph.py
index 4ec0cac1..332f198c 100644
--- a/reme/components/file_graph/neo4j_file_graph.py
+++ b/reme/components/file_graph/neo4j_file_graph.py
@@ -7,7 +7,7 @@ Property-graph mapping:
     Virtual node: (:File {path})  — placeholder created when something
                   links to a path that hasn't been upserted yet.
 
-    Edge:         (:File)-[:LINKS {idx, anchor, predicate}]->(:File)
+    Edge:         (:File)-[:LINKS {idx, anchor}]->(:File)
 
 The ``links_json`` property doubles as the "is real" marker — its
 presence means the node was upserted with a payload; its absence
@@ -43,7 +43,7 @@ from ...schema import FileLink, FileNode
 from ...schema.file_node import FileFrontMatter
 
 _TYPED_FRONTMATTER_FIELDS = {"name", "description"}
-_LINK_FIELDS = {"source_path", "target_path", "target_anchor", "predicate"}
+_LINK_FIELDS = {"source_path", "target_path", "target_anchor"}
 
 # Properties that distinguish a "real" node from a virtual placeholder.
 # Listed for the demote query (delete_nodes) so we can REMOVE them all.
@@ -169,7 +169,6 @@ class Neo4jFileGraph(BaseFileGraph):
                     {
                         "idx": i,
                         "anchor": link.target_anchor,
-                        "predicate": link.predicate,
                         "target": link.target_path,
                     }
                     for i, link in enumerate(node.links)
@@ -211,7 +210,7 @@ class Neo4jFileGraph(BaseFileGraph):
             UNWIND n.links AS link
             MERGE (t:File {path: link.target})
             MERGE (s)-[r:LINKS {idx: link.idx}]->(t)
-            SET r.anchor = link.anchor, r.predicate = link.predicate
+            SET r.anchor = link.anchor
             """,
             items=payload,
         )
@@ -313,7 +312,6 @@ class Neo4jFileGraph(BaseFileGraph):
                 {
                     "idx": i,
                     "anchor": link.get("target_anchor"),
-                    "predicate": link.get("predicate"),
                     "target": link.get("target_path"),
                 }
                 for i, link in enumerate(links)
@@ -340,7 +338,7 @@ class Neo4jFileGraph(BaseFileGraph):
             UNWIND n.links AS link
             MERGE (t:File {path: link.target})
             MERGE (s)-[r:LINKS {idx: link.idx}]->(t)
-            SET r.anchor = link.anchor, r.predicate = link.predicate
+            SET r.anchor = link.anchor
             """,
             items=payload,
         )
@@ -369,8 +367,7 @@ class Neo4jFileGraph(BaseFileGraph):
                 WHERE s.links_json IS NOT NULL
                 MATCH (s)-[r:LINKS]->(t:File)
                 WHERE 1=1 {target_filter}
-                RETURN t.path AS target, r.anchor AS anchor,
-                       r.predicate AS predicate, r.idx AS idx
+                RETURN t.path AS target, r.anchor AS anchor, r.idx AS idx
                 ORDER BY r.idx ASC
                 """,
                 path=path,
@@ -381,7 +378,6 @@ class Neo4jFileGraph(BaseFileGraph):
                 source_path=path,
                 target_path=row["target"],
                 target_anchor=row.get("anchor"),
-                predicate=row.get("predicate"),
             )
             for row in rows
         ]
@@ -405,8 +401,7 @@ class Neo4jFileGraph(BaseFileGraph):
                 WHERE 1=1 {target_filter}
                 MATCH (s:File)-[r:LINKS]->(t)
                 WHERE s.links_json IS NOT NULL
-                RETURN r.anchor AS anchor, r.predicate AS predicate,
-                       r.idx AS idx, s.path AS source
+                RETURN r.anchor AS anchor, r.idx AS idx, s.path AS source
                 ORDER BY s.path ASC, r.idx ASC
                 """,
                 path=path,
@@ -417,7 +412,6 @@ class Neo4jFileGraph(BaseFileGraph):
                 source_path=row["source"],
                 target_path=path,
                 target_anchor=row.get("anchor"),
-                predicate=row.get("predicate"),
             )
             for row in rows
         ]
diff --git a/reme/schema/file_link.py b/reme/schema/file_link.py
index b029d5a9..72324d01 100644
--- a/reme/schema/file_link.py
+++ b/reme/schema/file_link.py
@@ -7,12 +7,14 @@ class FileLink(BaseModel):
     """file link
     [[target_path]]
     [[target_path#target_anchor]]
-    predicate:: [[target_*]]
-    [predicate:: [[target_*]]]
     """
 
     model_config = ConfigDict(extra="forbid")
     source_path: str = Field(default=..., description="source file path relative to working dir")
     target_path: str = Field(default=..., description="target file path relative to working dir")
-    target_anchor: str | None = Field(default=None, description="Heading or block anchor (text after '#')")
-    predicate: str | None = Field(default=None, description="Dataview-style typed-link predicate")
+    target_anchor: str | None = Field(default=None, description="Heading, block, or line anchor (text after '#')")
+    predicate: str | None = Field(
+        default=None,
+        exclude=True,
+        description="Deprecated compatibility field; accepted when loading legacy indexes but otherwise unused",
+    )
diff --git a/reme/steps/evolve/dream/integrate.yaml b/reme/steps/evolve/dream/integrate.yaml
index 0189923d..6b52cb46 100644
--- a/reme/steps/evolve/dream/integrate.yaml
+++ b/reme/steps/evolve/dream/integrate.yaml
@@ -18,8 +18,8 @@ integrate_system_prompt_procedure: |
     genuinely needs more.
   - If the draft starts copying paragraphs or narrating the event, it belongs
     in the source material, not digest.
-  - Provenance edges carry details: cite every relevant unit_paths entry with
-    `derived_from:: [[<path>]]`.
+  - The Sources section carries details: cite every relevant unit_paths entry
+    as `- [[<path>]]`.
   - Digest-to-digest wikilinks carry the conceptual graph.
 
   ## Procedure Body Shape
@@ -30,9 +30,9 @@ integrate_system_prompt_procedure: |
   - Steps: numbered or terse bullets; each step is verb-led.
   - Pre-conditions / inputs: short list, not prose.
   - Failure modes / caveats: brief.
-  - `derived_from:: [[<material-path>]]`: at least one, and normally every
-    relevant path in unit_paths. Plain-prose provenance does NOT count; only
-    wikilinks survive future updates.
+  - `## Sources`: at least one `- [[<material-path>]]` item, and normally one
+    for every relevant path in unit_paths. Plain-prose provenance does NOT
+    count; only wikilinks survive future updates.
 
   ## Workflow
 
@@ -50,7 +50,7 @@ integrate_system_prompt_procedure: |
      - unrelated: ignore.
   3. Choose exactly one action:
      - CREATE: no same_abstraction hit; write `{digest_dir}/procedure/<slug>.md`.
-     - CORROBORATE: same procedure observed again; append derived_from and
+     - CORROBORATE: same procedure observed again; append its source link and
        optionally strengthen wording.
      - REFINE: new pre-condition, edge case, failure mode, scope, or step;
        expand the relevant span or slot the step into the right position.
@@ -58,21 +58,18 @@ integrate_system_prompt_procedure: |
        tighten or annotate inline with `> note: contradicted by [[<path>]] - <one-line>`.
   4. Weave related digest nodes into the body as wikilinks on both CREATE and
      UPDATE. UPDATE must be additive: never remove existing wikilinks or
-     derived_from entries. Default to weaving more, not less; this is the only
+     source links. Default to weaving more, not less; this is the only
      chance to attach recalled related nodes.
 
   ## Wikilink Graph
 
-  - Source provenance links point from digest back to material:
-    `derived_from:: [[daily/<date>/<session>.md]]` or
-    `derived_from:: [[resource/<path>]]`.
+  - A `## Sources` section points from digest back to material with bare list
+    items such as `- [[daily/<date>/<session>.md]]` and
+    `- [[resource/<path>]]`.
   - Procedure nodes may link to any digest bucket:
     `[[{digest_dir}/procedure/<slug>.md]]`,
     `[[{digest_dir}/personal/<slug>.md]]`, or
     `[[{digest_dir}/wiki/<slug>.md]]`.
-  - Useful predicates include `derived_from::`, `relates_to::`,
-    `depends_on::`, and `blocks_on::`; predicates are open and live outside
-    the brackets.
   - Wikilinks must be full workspace-relative paths with `.md`.
   - CREATE writes inside `{digest_dir}/procedure/`. UPDATE may target any
     bucket if recall legitimately found the same abstraction.
@@ -101,7 +98,7 @@ integrate_system_prompt_procedure_zh: |
 
   - 正文短且抽象，通常 50-200 words；只有流程本身确实需要时才更长。
   - 如果草稿开始复制段落或叙述事件，说明细节放错层了。
-  - Provenance edge 承载细节：用 `derived_from:: [[<path>]]` 引用 unit_paths 中每个相关来源。
+  - Sources 章节承载细节：用 `- [[<path>]]` 引用 unit_paths 中每个相关来源。
   - Digest 之间的 wikilink 承载概念图。
 
   ## Procedure 正文形态
@@ -112,7 +109,7 @@ integrate_system_prompt_procedure_zh: |
   - Steps：编号或短 bullet；每步以动词开头。
   - Pre-conditions / inputs：短列表，不写长 prose。
   - Failure modes / caveats：简短。
-  - `derived_from:: [[<material-path>]]`：至少一条，通常覆盖 unit_paths 中每个相关 path。
+  - `## Sources`：至少一条 `- [[<material-path>]]`，通常覆盖 unit_paths 中每个相关 path。
     纯文本 provenance 不算；只有 wikilink 会在未来更新中保留下来。
 
   ## 工作流
@@ -126,22 +123,20 @@ integrate_system_prompt_procedure_zh: |
      - unrelated：忽略。
   3. 选择且只选择一个 action：
      - CREATE：没有 same_abstraction 命中；写入 `{digest_dir}/procedure/<slug>.md`。
-     - CORROBORATE：同一流程再次出现；追加 derived_from，可选强化措辞。
+     - CORROBORATE：同一流程再次出现；追加来源链接，可选强化措辞。
      - REFINE：新增前置条件、边界情况、失败模式、适用范围或步骤；扩展相关段落或把步骤插到正确位置。
      - CORRECT：顺序错误、缺关键步骤、结果不好或有冲突；收紧表述，或用
        `> note: contradicted by [[<path>]] - <one-line>` 内联标注。
   4. CREATE 和 UPDATE 都要把 related digest 节点织入正文 wikilink。UPDATE 必须只增不删：不要删除已有
-     wikilink 或 derived_from。默认多织，而不是少织；这是挂接召回到的相关节点的唯一机会。
+     wikilink 或来源链接。默认多织，而不是少织；这是挂接召回到的相关节点的唯一机会。
 
   ## Wikilink 图
 
-  - 来源 provenance 从 digest 指回材料：
-    `derived_from:: [[daily/<date>/<session>.md]]` 或 `derived_from:: [[resource/<path>]]`。
+  - `## Sources` 章节用普通列表 Wikilink 从 digest 指回材料，例如
+    `- [[daily/<date>/<session>.md]]` 和 `- [[resource/<path>]]`。
   - Procedure 节点可以链接任意 digest bucket：
     `[[{digest_dir}/procedure/<slug>.md]]`、`[[{digest_dir}/personal/<slug>.md]]`、
     `[[{digest_dir}/wiki/<slug>.md]]`。
-  - 常用 predicate：`derived_from::`、`relates_to::`、`depends_on::`、`blocks_on::`；
-    predicate 词表开放，写在方括号外。
   - Wikilink 必须是带 `.md` 的 workspace-relative path。
   - CREATE 写入 `{digest_dir}/procedure/`。如果 recall 合法命中同一抽象，UPDATE 可以跨 bucket。
   - `edit` 只改正文，并尽量只增不删：不要删除 old span 中已有的 wikilink。
@@ -171,8 +166,8 @@ integrate_system_prompt_personal: |
 
   - Body is short and operational, usually 50-200 words.
   - Do not narrate what the user said in detail; cite source material instead.
-  - Provenance edges carry details: cite every relevant unit_paths entry with
-    `derived_from:: [[<path>]]`.
+  - The Sources section carries details: cite every relevant unit_paths entry
+    as `- [[<path>]]`.
   - Digest-to-digest wikilinks carry the conceptual graph.
 
   ## Personal Body Shape
@@ -185,9 +180,9 @@ integrate_system_prompt_personal: |
   - `How to apply:` contexts, tasks, boundaries, or exceptions.
   - Do not invent exceptions or soften a hard preference unless the source
     material explicitly supports that exception.
-  - `derived_from:: [[<material-path>]]`: at least one, and normally every
-    relevant path in unit_paths. Plain-prose provenance does NOT count; only
-    wikilinks survive future updates.
+  - `## Sources`: at least one `- [[<material-path>]]` item, and normally one
+    for every relevant path in unit_paths. Plain-prose provenance does NOT
+    count; only wikilinks survive future updates.
 
   For preferences, prefer one node per preference rather than one large person
   node; that is the granularity downstream search will hit.
@@ -208,7 +203,7 @@ integrate_system_prompt_personal: |
      - unrelated: ignore.
   3. Choose exactly one action:
      - CREATE: no same_abstraction hit; write `{digest_dir}/personal/<slug>.md`.
-     - CORROBORATE: the rule is reaffirmed; append derived_from and optionally
+     - CORROBORATE: the rule is reaffirmed; append its source link and optionally
        strengthen confidence.
      - REFINE: scope, condition, exception, or example changed; expand `How to apply:`.
      - CORRECT: the user/team changed their mind or evidence conflicts; tighten
@@ -216,21 +211,18 @@ integrate_system_prompt_personal: |
        `> note: contradicted by [[<path>]] - <one-line>`.
   4. Weave related digest nodes into the body as wikilinks on both CREATE and
      UPDATE. UPDATE must be additive: never remove existing wikilinks or
-     derived_from entries. Default to weaving more, not less; this is the only
+     source links. Default to weaving more, not less; this is the only
      chance to attach recalled related nodes.
 
   ## Wikilink Graph
 
-  - Source provenance links point from digest back to material:
-    `derived_from:: [[daily/<date>/<session>.md]]` or
-    `derived_from:: [[resource/<path>]]`.
+  - A `## Sources` section points from digest back to material with bare list
+    items such as `- [[daily/<date>/<session>.md]]` and
+    `- [[resource/<path>]]`.
   - Personal nodes may link to any digest bucket:
     `[[{digest_dir}/personal/<slug>.md]]`,
     `[[{digest_dir}/procedure/<slug>.md]]`, or
     `[[{digest_dir}/wiki/<slug>.md]]`.
-  - Useful predicates include `derived_from::`, `applies_to::`,
-    `relates_to::`, `depends_on::`, and `contradicts::`; predicates are open
-    and live outside the brackets.
   - Wikilinks must be full workspace-relative paths with `.md`.
   - CREATE writes inside `{digest_dir}/personal/`. UPDATE may target any bucket
     if recall legitimately found the same abstraction.
@@ -260,7 +252,7 @@ integrate_system_prompt_personal_zh: |
 
   - 正文短且可操作，通常 50-200 words。
   - 不要详细复述用户说了什么；用来源材料承载细节。
-  - Provenance edge 承载细节：用 `derived_from:: [[<path>]]` 引用 unit_paths 中每个相关来源。
+  - Sources 章节承载细节：用 `- [[<path>]]` 引用 unit_paths 中每个相关来源。
   - Digest 之间的 wikilink 承载概念图。
 
   ## Personal 正文形态
@@ -271,7 +263,7 @@ integrate_system_prompt_personal_zh: |
   - `Why:` 原因或上下文，帮助未来判断边界情况。
   - `How to apply:` 适用上下文、任务、边界或例外。
   - 不要凭空添加例外，也不要软化明确偏好；只有来源材料明确支持时才写例外。
-  - `derived_from:: [[<material-path>]]`：至少一条，通常覆盖 unit_paths 中每个相关 path。
+  - `## Sources`：至少一条 `- [[<material-path>]]`，通常覆盖 unit_paths 中每个相关 path。
     纯文本 provenance 不算；只有 wikilink 会在未来更新中保留下来。
 
   偏好类内容优先一条偏好一个 node，而不是一个人一个大 node；这是下游搜索更容易命中的粒度。
@@ -287,22 +279,20 @@ integrate_system_prompt_personal_zh: |
      - unrelated：忽略。
   3. 选择且只选择一个 action：
      - CREATE：没有 same_abstraction 命中；写入 `{digest_dir}/personal/<slug>.md`。
-     - CORROBORATE：规则被再次确认；追加 derived_from，可选强化置信度。
+     - CORROBORATE：规则被再次确认；追加来源链接，可选强化置信度。
      - REFINE：scope、条件、例外或例子变化；扩展 `How to apply:`。
      - CORRECT：用户/团队改变主意或证据冲突；收紧到新旧证据都支持的表述，或用
        `> note: contradicted by [[<path>]] - <one-line>` 内联标注。
   4. CREATE 和 UPDATE 都要把 related digest 节点织入正文 wikilink。UPDATE 必须只增不删：不要删除已有
-     wikilink 或 derived_from。默认多织，而不是少织；这是挂接召回到的相关节点的唯一机会。
+     wikilink 或来源链接。默认多织，而不是少织；这是挂接召回到的相关节点的唯一机会。
 
   ## Wikilink 图
 
-  - 来源 provenance 从 digest 指回材料：
-    `derived_from:: [[daily/<date>/<session>.md]]` 或 `derived_from:: [[resource/<path>]]`。
+  - `## Sources` 章节用普通列表 Wikilink 从 digest 指回材料，例如
+    `- [[daily/<date>/<session>.md]]` 和 `- [[resource/<path>]]`。
   - Personal 节点可以链接任意 digest bucket：
     `[[{digest_dir}/personal/<slug>.md]]`、`[[{digest_dir}/procedure/<slug>.md]]`、
     `[[{digest_dir}/wiki/<slug>.md]]`。
-  - 常用 predicate：`derived_from::`、`applies_to::`、`relates_to::`、`depends_on::`、
-    `contradicts::`；predicate 词表开放，写在方括号外。
   - Wikilink 必须是带 `.md` 的 workspace-relative path。
   - CREATE 写入 `{digest_dir}/personal/`。如果 recall 合法命中同一抽象，UPDATE 可以跨 bucket。
   - `edit` 只改正文，并尽量只增不删：不要删除 old span 中已有的 wikilink。
@@ -335,8 +325,8 @@ integrate_system_prompt_wiki: |
     genuinely needs more.
   - If the draft starts copying paragraphs or narrating the event, it belongs
     in the source material, not digest.
-  - Provenance edges carry details: cite every relevant unit_paths entry with
-    `derived_from:: [[<path>]]`.
+  - The Sources section carries details: cite every relevant unit_paths entry
+    as `- [[<path>]]`.
   - Digest-to-digest wikilinks carry the conceptual graph.
 
   ## Wiki Body Shape
@@ -346,11 +336,10 @@ integrate_system_prompt_wiki: |
   - First line: one-sentence definition or claim.
   - Body: short paragraphs or tight bullets with properties, sub-claims,
     distinctions, and one-line examples.
-  - Relations: typed wikilinks where the relation has semantic weight; most
-    cross-node links can stay bare.
-  - `derived_from:: [[<material-path>]]`: at least one, and normally every
-    relevant path in unit_paths. Plain-prose provenance does NOT count; only
-    wikilinks survive future updates.
+  - Relations: weave related nodes into natural prose with ordinary wikilinks.
+  - `## Sources`: at least one `- [[<material-path>]]` item, and normally one
+    for every relevant path in unit_paths. Plain-prose provenance does NOT
+    count; only wikilinks survive future updates.
 
   ## Workflow
 
@@ -368,8 +357,8 @@ integrate_system_prompt_wiki: |
      - unrelated: ignore.
   3. Choose exactly one action:
      - CREATE: no same_abstraction hit; write `{digest_dir}/wiki/<slug>.md`.
-     - CORROBORATE: the principle is reaffirmed by a new instance; append
-       derived_from and optionally strengthen wording.
+     - CORROBORATE: the principle is reaffirmed by a new instance; append its
+       source link and optionally strengthen wording.
      - REFINE: nuance, scope, edge case, or framing changes; tighten the
        relevant span. Body grows in precision, not detail volume.
      - CORRECT: factual contradiction or overstatement; tighten to the narrower
@@ -377,22 +366,18 @@ integrate_system_prompt_wiki: |
        `> note: contradicted by [[<path>]] - <one-line>`.
   4. Weave related digest nodes into the body as wikilinks on both CREATE and
      UPDATE. UPDATE must be additive: never remove existing wikilinks or
-     derived_from entries. Default to weaving more, not less; this is the only
+     source links. Default to weaving more, not less; this is the only
      chance to attach recalled related nodes.
 
   ## Wikilink Graph
 
-  - Source provenance links point from digest back to material:
-    `derived_from:: [[daily/<date>/<session>.md]]` or
-    `derived_from:: [[resource/<path>]]`.
+  - A `## Sources` section points from digest back to material with bare list
+    items such as `- [[daily/<date>/<session>.md]]` and
+    `- [[resource/<path>]]`.
   - Wiki nodes may link to any digest bucket:
     `[[{digest_dir}/wiki/<slug>.md]]`,
     `[[{digest_dir}/procedure/<slug>.md]]`, or
     `[[{digest_dir}/personal/<slug>.md]]`.
-  - Useful predicates include `derived_from::`, `is_a::`, `extends::`,
-    `depends_on::`, `relates_to::`, and `contradicts::`; predicates are open
-    and live outside the brackets. Most cross-node links can stay bare unless
-    the relation has clear semantic weight.
   - Wikilinks must be full workspace-relative paths with `.md`.
   - CREATE writes inside `{digest_dir}/wiki/`. UPDATE may target any bucket if
     recall legitimately found the same abstraction.
@@ -422,7 +407,7 @@ integrate_system_prompt_wiki_zh: |
 
   - 正文短且抽象，通常 50-200 words；只有概念本身确实需要时才更长。
   - 如果草稿开始复制段落或叙述事件，说明细节放错层了。
-  - Provenance edge 承载细节：用 `derived_from:: [[<path>]]` 引用 unit_paths 中每个相关来源。
+  - Sources 章节承载细节：用 `- [[<path>]]` 引用 unit_paths 中每个相关来源。
   - Digest 之间的 wikilink 承载概念图。
 
   ## Wiki 正文形态
@@ -431,8 +416,8 @@ integrate_system_prompt_wiki_zh: |
 
   - First line：一句话定义或主张。
   - Body：短段落或紧凑 bullets，写属性、子主张、区分和一行例子。
-  - Relations：有明确语义重量时使用 typed wikilink；大多数 cross-node link 可以裸写。
-  - `derived_from:: [[<material-path>]]`：至少一条，通常覆盖 unit_paths 中每个相关 path。
+  - Relations：在自然语言中用普通 Wikilink 织入相关节点。
+  - `## Sources`：至少一条 `- [[<material-path>]]`，通常覆盖 unit_paths 中每个相关 path。
     纯文本 provenance 不算；只有 wikilink 会在未来更新中保留下来。
 
   ## 工作流
@@ -446,23 +431,20 @@ integrate_system_prompt_wiki_zh: |
      - unrelated：忽略。
   3. 选择且只选择一个 action：
      - CREATE：没有 same_abstraction 命中；写入 `{digest_dir}/wiki/<slug>.md`。
-     - CORROBORATE：原则被新实例再次确认；追加 derived_from，可选强化措辞。
+     - CORROBORATE：原则被新实例再次确认；追加来源链接，可选强化措辞。
      - REFINE：nuance、scope、edge case 或 framing 改变；收紧相关段落。正文增长的是精度，不是细节量。
      - CORRECT：事实冲突或过度概括；收紧到更窄且有支持的表述，或用
        `> note: contradicted by [[<path>]] - <one-line>` 内联标注。
   4. CREATE 和 UPDATE 都要把 related digest 节点织入正文 wikilink。UPDATE 必须只增不删：不要删除已有
-     wikilink 或 derived_from。默认多织，而不是少织；这是挂接召回到的相关节点的唯一机会。
+     wikilink 或来源链接。默认多织，而不是少织；这是挂接召回到的相关节点的唯一机会。
 
   ## Wikilink 图
 
-  - 来源 provenance 从 digest 指回材料：
-    `derived_from:: [[daily/<date>/<session>.md]]` 或 `derived_from:: [[resource/<path>]]`。
+  - `## Sources` 章节用普通列表 Wikilink 从 digest 指回材料，例如
+    `- [[daily/<date>/<session>.md]]` 和 `- [[resource/<path>]]`。
   - Wiki 节点可以链接任意 digest bucket：
     `[[{digest_dir}/wiki/<slug>.md]]`、`[[{digest_dir}/procedure/<slug>.md]]`、
     `[[{digest_dir}/personal/<slug>.md]]`。
-  - 常用 predicate：`derived_from::`、`is_a::`、`extends::`、`depends_on::`、
-    `relates_to::`、`contradicts::`；predicate 词表开放，写在方括号外。有明确语义重量时才加
-    predicate，大多数 cross-node link 可以裸写。
   - Wikilink 必须是带 `.md` 的 workspace-relative path。
   - CREATE 写入 `{digest_dir}/wiki/`。如果 recall 合法命中同一抽象，UPDATE 可以跨 bucket。
   - `edit` 只改正文，并尽量只增不删：不要删除 old span 中已有的 wikilink。
@@ -489,8 +471,8 @@ integrate_user_message: |
   {material_blob}
 
   Integrate this single unit into digest memory. Cite every relevant unit_paths
-  entry with `derived_from:: [[<path>]]`, recall related digest nodes, and weave
-  useful digest wikilinks into the target node.
+  entry as `- [[<path>]]` under `## Sources`, recall related digest nodes, and
+  weave useful digest wikilinks into the target node.
 
 integrate_user_message_zh: |
   提示：{hint}
@@ -506,5 +488,5 @@ integrate_user_message_zh: |
 
   {material_blob}
 
-  将这个单独的 unit 整合进 digest memory。用 `derived_from:: [[<path>]]` 引用 unit_paths 中每个相关来源，
-  召回相关 digest 节点，并把有用的 digest wikilink 织入目标节点。
+  将这个单独的 unit 整合进 digest memory。在 `## Sources` 下用 `- [[<path>]]` 引用 unit_paths 中每个
+  相关来源，召回相关 digest 节点，并把有用的 digest wikilink 织入目标节点。
diff --git a/reme/steps/evolve/dream/utils.py b/reme/steps/evolve/dream/utils.py
index b731ff18..b6e7edc5 100644
--- a/reme/steps/evolve/dream/utils.py
+++ b/reme/steps/evolve/dream/utils.py
@@ -71,10 +71,10 @@ def scan_day_files(workspace: Path, day: str, daily: str, interests_name: str =
     out: list[str] = []
     day_index = workspace / daily / f"{day}.md"
     if day_index.is_file():
-        out.append(str(day_index.relative_to(workspace)))
+        out.append(day_index.relative_to(workspace).as_posix())
     daily_root = workspace / daily / day
     if daily_root.is_dir():
-        out.extend(str(p.relative_to(workspace)) for p in sorted(daily_root.rglob("*.md")) if p.is_file())
+        out.extend(p.relative_to(workspace).as_posix() for p in sorted(daily_root.rglob("*.md")) if p.is_file())
     return [p for p in out if p != f"{daily}/{day}/{interests_name}"]
 
 
diff --git a/reme/steps/file_io/move.py b/reme/steps/file_io/move.py
index c9e4ab43..e1bad6ba 100644
--- a/reme/steps/file_io/move.py
+++ b/reme/steps/file_io/move.py
@@ -65,6 +65,7 @@ class MoveStep(BaseStep):
             )
         self.context.response.metadata.update(payload)
 
+    # pylint: disable=too-many-return-statements
     async def _move(self, src_path: str, dst_path: str, overwrite: bool, retarget: bool) -> dict:
         workspace_dir = Path(self.file_store.workspace_path or ".").resolve()
         src_abs, src_err = resolve_path(workspace_dir, src_path) if src_path else (None, "src_path is required")
@@ -79,7 +80,8 @@ class MoveStep(BaseStep):
         assert src_abs is not None and dst_abs is not None  # narrowed by precheck
         dst_abs.parent.mkdir(parents=True, exist_ok=True)
 
-        # Step 1 — copy. Both files exist; inbound [[src_path]] still resolves.
+        # Step 1 — copy. Wikilinks use workspace-relative paths, so outgoing
+        # targets do not change when their containing document moves.
         shutil.copyfile(str(src_abs), str(dst_abs))
         payload: dict = {"src_path": src_path, "dst_path": dst_path, "size": dst_abs.stat().st_size}
 
@@ -91,10 +93,13 @@ class MoveStep(BaseStep):
             try:
                 report = await WikilinkHandler.retarget_links(self.file_store, src=src_path, dst=dst_path)
             except Exception as exc:
-                payload["retarget"] = {"error": f"retarget raised: {exc!r}"}
+                error = f"retarget raised: {exc!r}"
+                payload["error"] = error
+                payload["retarget"] = {"error": error}
                 payload["src_removed"] = False
                 return payload
             if "error" in report:
+                payload["error"] = report["error"]
                 payload["retarget"] = report
                 payload["src_removed"] = False
                 return payload
diff --git a/reme/steps/file_io/read.py b/reme/steps/file_io/read.py
index 56fcecbe..ce530a87 100644
--- a/reme/steps/file_io/read.py
+++ b/reme/steps/file_io/read.py
@@ -168,7 +168,7 @@ class ReadStep(BaseStep):
         """Append the rendered neighbor block + stash raw expansion in metadata."""
         assert self.context is not None
         try:
-            rel_path = str(target.relative_to(self.workspace_path))
+            rel_path = target.relative_to(self.workspace_path.resolve()).as_posix()
         except ValueError:
             self.logger.info(f"[{self.name}] skip neighbors: path outside workspace_path path={target}")
             return
diff --git a/reme/steps/index/traverse.py b/reme/steps/index/traverse.py
index 98665977..93d4134b 100644
--- a/reme/steps/index/traverse.py
+++ b/reme/steps/index/traverse.py
@@ -1,10 +1,10 @@
 """BFS over wikilink edges from one or more seed files.
 
 One record per traversed *edge* (not per node): the same target can repeat
-if reached via different predicates or paths. Each record carries the
-predecessor plus the link's predicate/anchor so callers can reconstruct
-the path. Adjacency is built once via a single ``file_store.get_nodes()``
-call — BFS then runs purely in memory with no per-frontier round-trips.
+if reached via different anchors or paths. Each record carries the
+predecessor and anchor so callers can reconstruct the path. Adjacency is
+built once via a single ``file_store.get_nodes()`` call — BFS then runs
+purely in memory with no per-frontier round-trips.
 """
 
 from collections import deque
@@ -46,7 +46,7 @@ def _bfs(
     outbound: Adjacency,
     inbound: Adjacency,
 ) -> list[dict]:
-    """In-memory BFS; emits one record per unique (src, dst, predicate) edge."""
+    """In-memory BFS; emits one record per unique (src, dst, anchor) edge."""
     sources: list[Adjacency] = []
     if direction in _OUT:
         sources.append(outbound)
@@ -63,7 +63,7 @@ def _bfs(
             continue
         for src in sources:
             for next_path, link in src.get(current, ()):
-                key = (current, next_path, link.predicate)
+                key = (current, next_path, link.target_anchor)
                 if key in visited:
                     continue
                 visited.add(key)
@@ -72,7 +72,6 @@ def _bfs(
                         "path": next_path,
                         "depth": depth + 1,
                         "via": current,
-                        "predicate": link.predicate,
                         "anchor": link.target_anchor,
                     },
                 )
@@ -119,8 +118,7 @@ class TraverseStep(BaseStep):
                 target = r["path"]
                 if r["anchor"]:
                     target = f"{target}#{r['anchor']}"
-                predicate = r["predicate"] or "-"
-                lines.append(f"[depth={r['depth']}] {r['via']} --{predicate}--> {target}")
+                lines.append(f"[depth={r['depth']}] {r['via']} --> {target}")
             answer = "\n".join(lines)
 
         self.context.response.success = True
diff --git a/reme/utils/__init__.py b/reme/utils/__init__.py
index a41e66e9..c9b7aaf3 100644
--- a/reme/utils/__init__.py
+++ b/reme/utils/__init__.py
@@ -9,6 +9,7 @@ from .common_utils import (
 )
 from .env_utils import load_env, parse_env_file
 from .link_expansion import expand_links, render_expansion_lines
+from .line_anchor import format_line_anchor, parse_line_anchor
 from .logger_utils import get_logger
 from .logo_utils import print_logo
 from .service_utils import find_reme, locate_reme, precheck_start, cli_find_reme, running_service_config
@@ -33,6 +34,8 @@ __all__ = [
     "parse_env_file",
     "expand_links",
     "render_expansion_lines",
+    "format_line_anchor",
+    "parse_line_anchor",
     "get_logger",
     "print_logo",
     "find_reme",
diff --git a/reme/utils/line_anchor.py b/reme/utils/line_anchor.py
new file mode 100644
index 00000000..414eaed1
--- /dev/null
+++ b/reme/utils/line_anchor.py
@@ -0,0 +1,43 @@
+"""Parse and format GitHub-style 1-based line-anchor strings."""
+
+import re
+
+_LINE_ANCHOR_RE = re.compile(r"L[0-9]+(?:-L[0-9]+)?(?:,L[0-9]+(?:-L[0-9]+)?)*")
+
+
+def parse_line_anchor(anchor: str | None) -> list[tuple[int, int]] | None:
+    """Return normalized inclusive ranges, or ``None`` for a non-line anchor.
+
+    Supported forms are ``L9``, ``L9-L10`` and
+    ``L9-L10,L15-L20``. Overlapping and adjacent ranges are merged.
+    Anchors beginning with ``L<digit>`` are treated as line anchors and raise
+    ``ValueError`` when malformed, zero-based, or reversed.
+    """
+    if not anchor or not re.match(r"L[0-9]", anchor):
+        return None
+    if not _LINE_ANCHOR_RE.fullmatch(anchor):
+        raise ValueError(f"invalid line anchor: #{anchor}")
+
+    ranges: list[tuple[int, int]] = []
+    for item in anchor.split(","):
+        start_text, separator, end_text = item.partition("-L")
+        start = int(start_text[1:])
+        end = int(end_text) if separator else start
+        if start < 1 or end < 1:
+            raise ValueError("line numbers must be at least 1")
+        if start > end:
+            raise ValueError(f"line range start ({start}) exceeds end ({end})")
+        ranges.append((start, end))
+
+    merged: list[tuple[int, int]] = []
+    for start, end in sorted(ranges):
+        if merged and start <= merged[-1][1] + 1:
+            merged[-1] = (merged[-1][0], max(merged[-1][1], end))
+        else:
+            merged.append((start, end))
+    return merged
+
+
+def format_line_anchor(ranges: list[tuple[int, int]]) -> str:
+    """Render normalized ranges without the leading ``#``."""
+    return ",".join(f"L{start}" if start == end else f"L{start}-L{end}" for start, end in ranges)
diff --git a/reme/utils/link_expansion.py b/reme/utils/link_expansion.py
index f3920c67..5ec759e2 100644
--- a/reme/utils/link_expansion.py
+++ b/reme/utils/link_expansion.py
@@ -8,7 +8,7 @@ Two-layer split so callers can pick what they need:
 
 * :func:`expand_links` — data layer. Returns a structured dict keyed
   by source path, each value carrying its outlinks / inlinks with
-  neighbor meta and per-edge predicate/anchor.
+  neighbor metadata and link anchors.
 * :func:`render_expansion_lines` — view layer. Turns one path's
   expansion sub-dict into the same ``  → path  name=… description=…``
   block ``SearchStep`` has historically printed.
@@ -19,16 +19,16 @@ import asyncio
 from ..schema import FileLink, FileNode
 
 
-def _group_by_neighbor(links: list[FileLink], key_attr: str) -> dict[str, list[dict]]:
-    """Group edges by neighbor path (insertion-ordered), each value a list of {predicate, anchor}."""
-    out: dict[str, list[dict]] = {}
+def _group_by_neighbor(links: list[FileLink], key_attr: str) -> dict[str, list[str]]:
+    """Group anchors by neighbor path while preserving insertion order."""
+    out: dict[str, list[str]] = {}
     for lnk in links:
         neighbor = getattr(lnk, key_attr)
         if not neighbor:
             continue
-        out.setdefault(neighbor, []).append(
-            {"predicate": lnk.predicate, "anchor": lnk.target_anchor},
-        )
+        anchors = out.setdefault(neighbor, [])
+        if lnk.target_anchor:
+            anchors.append(lnk.target_anchor)
     return out
 
 
@@ -55,16 +55,6 @@ def _format_meta_inline(meta: dict) -> str:
     return "  ".join(parts) if parts else "(no meta)"
 
 
-def _format_via(edge: dict) -> str:
-    """Render a single (predicate, anchor) edge as a 'via ...' descriptor."""
-    bits = []
-    if edge.get("predicate"):
-        bits.append(f"predicate={edge['predicate']}")
-    if edge.get("anchor"):
-        bits.append(f"anchor=#{edge['anchor']}")
-    return ", ".join(bits) if bits else "plain"
-
-
 async def expand_links(
     file_store,
     paths: list[str],
@@ -73,7 +63,7 @@ async def expand_links(
     """Fetch out/in links for each path and attach neighbor meta.
 
     Returns ``{path: {"outlinks": [...], "inlinks": [...]}, ...}`` where
-    each list item is ``{"path": str, "meta": {...}, "edges": [{"predicate", "anchor"}, ...]}``.
+    each list item is ``{"path": str, "meta": {...}, "anchors": [str, ...]}``.
     Empty input returns ``{}``. ``max_per_direction`` caps the neighbor
     list per direction *before* meta lookup so we don't fetch nodes
     that won't be displayed.
@@ -95,9 +85,10 @@ async def expand_links(
     nodes = await file_store.get_nodes(neighbor_paths) if neighbor_paths else []
     meta_by_path = {n.path: _node_meta(n) for n in nodes}
 
-    def _attach(grouped: dict[str, list[dict]]) -> list[dict]:
+    def _attach(grouped: dict[str, list[str]]) -> list[dict]:
         return [
-            {"path": npath, "meta": meta_by_path.get(npath, {}), "edges": edges} for npath, edges in grouped.items()
+            {"path": npath, "meta": meta_by_path.get(npath, {}), "anchors": anchors}
+            for npath, anchors in grouped.items()
         ]
 
     return {p: {"outlinks": _attach(og), "inlinks": _attach(ig)} for p, og, ig in zip(paths, out_grouped, in_grouped)}
@@ -124,6 +115,6 @@ def render_expansion_lines(expansion: dict, indent: str = "  ") -> list[str]:
         lines.append(f"{indent}{direction} ({len(items)}):")
         for item in items:
             lines.append(f"{inner}{arrow} {item['path']}  {_format_meta_inline(item['meta'])}")
-            for edge in item["edges"]:
-                lines.append(f"{edge_indent}via {_format_via(edge)}")
+            for anchor in item["anchors"]:
+                lines.append(f"{edge_indent}via anchor=#{anchor}")
     return lines
diff --git a/reme/utils/wikilink_handler.py b/reme/utils/wikilink_handler.py
index 9714aa16..8c17a2e0 100644
--- a/reme/utils/wikilink_handler.py
+++ b/reme/utils/wikilink_handler.py
@@ -2,7 +2,7 @@
 
 One class, :class:`WikilinkHandler`, owning every wikilink concern:
 
-* **Pure text** — regex, Dataview predicate inference, validation:
+* **Pure text** — regex, extraction, rewrite, validation:
   :meth:`~WikilinkHandler.extract_links` (used by
   :mod:`reme.components.file_chunker.markdown_file_chunker`),
   :meth:`~WikilinkHandler.scan_and_rewrite`,
@@ -16,13 +16,12 @@ One class, :class:`WikilinkHandler`, owning every wikilink concern:
   post-rename to point inbound ``[[src]]`` at the new path). Source
   candidates come from the file_graph's reverse index — no fs scan.
 
-Wikilink convention. Targets are taken **literally** — ``[[X]]`` →
+Wikilink targets are taken **literally** — ``[[X]]`` →
 ``target="X"``, no implicit ``.md``, no short-form basename search,
 no folder-note expansion. Anchor and alias survive a rewrite
-verbatim. Image marker (``!``) and Dataview predicate (``pred::``
-outside the brackets) sit outside ``[[...]]`` and are not touched by
-a rewrite. Recommended form: full path relative to the workspace with
-extension (``[[topics/x.md]]``).
+verbatim. Text outside ``[[...]]`` is ignored. Recommended form: full
+path relative to the workspace with extension (``[[topics/x.md]]``).
+Ordinary Markdown links such as ``[label](path.md)`` are ignored.
 
 Stale graph entries are harmless (``scan_and_rewrite`` returns
 count=0 and the file is skipped), but a graph missing recent writes
@@ -37,26 +36,23 @@ from ..enumeration import LinkScopeEnum
 from ..schema import FileLink
 
 
+def _normalize_workspace_path(path: str) -> str:
+    """Use POSIX separators for workspace paths on every platform."""
+    return path.replace("\\", "/")
+
+
 @dataclass(frozen=True)
 class WikilinkMatch:
-    """One ``[[...]]`` occurrence with parts surfaced.
-
-    ``anchor`` / ``alias`` are stored **without** the leading ``#`` /
-    ``|`` so they map cleanly to :class:`FileLink.target_anchor`; the
-    rewrite path reads the raw regex groups (with delimiters) directly
-    and doesn't go through this dataclass.
-    """
+    """The graph-relevant parts and source span of one wikilink."""
 
     target: str
     anchor: str | None
-    alias: str | None
-    bang: bool
     start: int
     end: int
 
 
 class WikilinkHandler:
-    """Pure-text wikilink operations: parse, extract, rewrite, validate."""
+    """Parse, extract, rewrite, and validate wikilinks."""
 
     # Captures: optional image marker (``!``), the bare target, an
     # optional ``#anchor`` slice (with ``#``), and an optional ``|alias``
@@ -77,54 +73,39 @@ class WikilinkHandler:
 
     FORBIDDEN_IN_NEW = ("[", "]", "#", "|", "\n", "\r")
 
-    _DATAVIEW_LINE_RE = re.compile(
-        r"^[ \t]*(?:[-*+][ \t]+)?(?P<predicate>[A-Za-z][A-Za-z0-9_]*)\s*::\s*(?P<value>.+?)\s*$",
-        re.MULTILINE,
-    )
-
-    _INLINE_FIELD_OPEN_RE = re.compile(r"\[(?P<predicate>[A-Za-z][A-Za-z0-9_]*)\s*::\s*")
-
     # -- Low-level scan ------------------------------------------------
 
     @classmethod
     def iter_matches(cls, text: str):
-        """Yield :class:`WikilinkMatch` for every ``[[...]]`` in ``text``.
-
-        Skips matches whose target is empty after strip (defensive).
-        """
+        """Yield every non-empty ``[[...]]`` occurrence in ``text``."""
         for m in cls.WIKILINK_RE.finditer(text):
             target = m.group("target").strip()
             if not target:
                 continue
             anchor_raw = m.group("anchor")
-            alias_raw = m.group("alias")
             yield WikilinkMatch(
                 target=target,
                 anchor=anchor_raw[1:].strip() if anchor_raw else None,
-                alias=alias_raw[1:].strip() if alias_raw else None,
-                bang=bool(m.group("bang")),
                 start=m.start(),
                 end=m.end(),
             )
 
-    # -- FileLink extraction (with predicate inference) ---------------
+    # -- FileLink extraction ------------------------------------------
 
     @classmethod
     def extract_links(cls, text: str, source_path: str) -> list[FileLink]:
         """Emit :class:`FileLink` edges for every wikilink in ``text``.
 
-        No resolution: ``target_path`` is the bracket contents verbatim.
-        Results are deduped by ``(target_path, predicate, target_anchor)``
-        preserving order.
+        Targets remain literal. Results are deduped by
+        ``(target_path, target_anchor)`` while preserving order.
         """
         if not text:
             return []
-        inline_spans = cls._iter_inline_fields(text)
+        source_path = _normalize_workspace_path(source_path)
         out: list[FileLink] = []
         seen: set[tuple] = set()
         for wm in cls.iter_matches(text):
-            predicate = cls._predicate_for(text, wm.start, inline_spans)
-            key = (wm.target, predicate, wm.anchor)
+            key = (wm.target, wm.anchor)
             if key in seen:
                 continue
             seen.add(key)
@@ -133,7 +114,6 @@ class WikilinkHandler:
                     source_path=source_path,
                     target_path=wm.target,
                     target_anchor=wm.anchor,
-                    predicate=predicate,
                 ),
             )
         return out
@@ -147,20 +127,19 @@ class WikilinkHandler:
         old: str,
         new: str | None,
     ) -> tuple[str, int]:
-        """Find (and optionally rewrite) wikilinks whose target equals ``old``.
+        """Find and optionally rewrite wikilinks whose target is ``old``.
 
         Returns ``(new_text, count)``. When ``new`` is ``None`` no rewrite
         happens (the original text is returned), but the count is still
-        populated — used by ``find_inbound``. Matching is literal:
-        ``target == old``. No short-link, no implicit ``.md``, no
-        folder-note expansion.
+        populated — used by ``find_inbound``. Matching is literal.
         """
+        old = _normalize_workspace_path(old)
+        new = _normalize_workspace_path(new) if new is not None else None
         count = 0
 
         def sub(match: re.Match) -> str:
             nonlocal count
-            target = match.group("target").strip()
-            if target != old:
+            if match.group("target").strip() != old:
                 return match.group(0)
             count += 1
             if new is None:
@@ -170,8 +149,7 @@ class WikilinkHandler:
             bang = match.group("bang") or ""
             return f"{bang}[[{new}{anchor}{alias}]]"
 
-        new_text = cls.WIKILINK_RE.sub(sub, text)
-        return new_text, count
+        return cls.WIKILINK_RE.sub(sub, text), count
 
     # -- Validation ----------------------------------------------------
 
@@ -201,53 +179,6 @@ class WikilinkHandler:
         prefix = scope.rstrip("/") + "/"
         return rel == scope or rel.startswith(prefix)
 
-    # -- Predicate helpers (internal) ---------------------------------
-
-    @classmethod
-    def _iter_inline_fields(cls, text: str) -> list[tuple[int, int, str]]:
-        """Find inline-bracketed ``[predicate:: …]`` field spans by depth scan."""
-        out: list[tuple[int, int, str]] = []
-        for m in cls._INLINE_FIELD_OPEN_RE.finditer(text):
-            depth = 1
-            i = m.end()
-            n = len(text)
-            while i < n:
-                c = text[i]
-                if c == "\n":
-                    break
-                if c == "[":
-                    depth += 1
-                elif c == "]":
-                    depth -= 1
-                    if depth == 0:
-                        out.append((m.start(), i + 1, m.group("predicate")))
-                        break
-                i += 1
-        return out
-
-    @classmethod
-    def _predicate_for(
-        cls,
-        text: str,
-        pos: int,
-        inline_spans: list[tuple[int, int, str]],
-    ) -> str | None:
-        """Resolve the predicate governing a wikilink at offset ``pos``.
-
-        Precedence: inline-bracketed > line-level Dataview > none.
-        """
-        for field_start, field_end, predicate in inline_spans:
-            if field_start <= pos < field_end:
-                return predicate
-        line_start = text.rfind("\n", 0, pos) + 1
-        line_end = text.find("\n", pos)
-        if line_end == -1:
-            line_end = len(text)
-        m = cls._DATAVIEW_LINE_RE.match(text[line_start:line_end])
-        if m and line_start + m.start("value") <= pos:
-            return m.group("predicate")
-        return None
-
     # -- Async file_graph-aware operations -----------------------------
 
     @classmethod
@@ -260,11 +191,12 @@ class WikilinkHandler:
         required to surface sources whose edges sit in the pending bucket.
         Each returned ``FileLink`` carries the linking node's ``source_path``;
         we dedupe to a sorted list since one source can host multiple edges
-        (different anchor/predicate) to the same target. Returns ``[]`` when
+        (different anchors) to the same target. Returns ``[]`` when
         there is no file_graph attached or no source references the target.
         """
         if not file_store.file_graph:
             return []
+        target = _normalize_workspace_path(target)
         inlinks = await file_store.file_graph.get_inlinks(target, scope=LinkScopeEnum.ALL)
         return sorted({link.source_path for link in inlinks if link.source_path})
 
@@ -272,12 +204,10 @@ class WikilinkHandler:
     async def find_inbound(cls, file_store, target: str, scope: str = "") -> dict:
         """Count wikilinks across the workspace that point at ``target``.
 
-        Literal matching: ``[[target]]`` only. The target file itself is
-        excluded — self-references don't survive a delete and aren't
-        actionable for the caller. Sources come from the file_graph's
-        reverse index; per-file counts come from reading each candidate
-        source (the graph dedupes by ``(target, predicate, anchor)`` so
-        it can't count repeated bare-wikilink occurrences directly).
+        The target file itself is excluded — self-references don't survive a
+        delete and aren't actionable for the caller. Sources come from the
+        file_graph's reverse index; per-file counts come from reading each
+        candidate source because the graph dedupes repeated edges.
 
         Result shape::
 
@@ -291,6 +221,8 @@ class WikilinkHandler:
 
         On bad inputs returns ``{"target": ..., "error": str}``.
         """
+        target = _normalize_workspace_path(target)
+        scope = _normalize_workspace_path(scope)
         if not target:
             return {"target": target, "error": "target is required"}
         if Path(target).is_absolute():
@@ -340,6 +272,9 @@ class WikilinkHandler:
         matching only; candidate sources come from the file_graph's reverse
         index.
         """
+        src = _normalize_workspace_path(src)
+        dst = _normalize_workspace_path(dst)
+        scope = _normalize_workspace_path(scope)
         err = cls.validate_src_dst(src, dst)
         if err is not None:
             return {"src": src, "dst": dst, "error": err}
diff --git a/tests/integration/_workspace_fixture.py b/tests/integration/_workspace_fixture.py
index 516610cc..b1d15641 100644
--- a/tests/integration/_workspace_fixture.py
+++ b/tests/integration/_workspace_fixture.py
@@ -84,7 +84,8 @@ token used to assert identity and claims between parties.
 ## Related
 Often issued by [[digest/wiki/oauth2.md]] flows.
 
-derived_from:: [[daily/2026-05-15/auth-design/notes.md]]
+## Sources
+- [[daily/2026-05-15/auth-design/notes.md]]
 """,
     "digest/wiki/oauth2.md": """\
 ---
@@ -103,7 +104,8 @@ issued by an authorization server.
 - Client credentials
 - Refresh token
 
-derived_from:: [[daily/2026-05-10/oauth-intro/notes.md]]
+## Sources
+- [[daily/2026-05-10/oauth-intro/notes.md]]
 """,
     "digest/procedure/key-rotation.md": """\
 ---
@@ -127,7 +129,8 @@ issuance.
 Default rotation cadence is **30 days**. Driven by historical practice;
 no formal compliance requirement has tightened this so far.
 
-derived_from:: [[daily/2026-05-20/rotation-plan/notes.md]]
+## Sources
+- [[daily/2026-05-20/rotation-plan/notes.md]]
 """,
     "digest/personal/no-trailing-summary.md": """\
 ---
@@ -144,7 +147,8 @@ description: 不要在回复末尾加总结段落
 **How to apply**: 任意编码 / 编辑任务回复结束时,直接停在最后一条
 有信息量的话上,不要再补一段"以上就是本次的修改..."。
 
-derived_from:: [[daily/2026-05-01/style-feedback/notes.md]]
+## Sources
+- [[daily/2026-05-01/style-feedback/notes.md]]
 """,
     # ----- daily provenance stubs (so the digest links don't dangle) -----
     "daily/2026-05-01/style-feedback/notes.md": """\
diff --git a/tests/integration/test_auto_dream.py b/tests/integration/test_auto_dream.py
index 2e271c2a..e020cf18 100644
--- a/tests/integration/test_auto_dream.py
+++ b/tests/integration/test_auto_dream.py
@@ -182,15 +182,13 @@ def test_auto_dream_and_proactive():
                     if (env.workspace_dir / rel).is_file()
                 }
                 digest_wikilinks = [rel for rel, text in target_texts.items() if "[[digest/" in text]
-                provenance_links = [
-                    rel for rel, text in target_texts.items() if f"derived_from:: [[{DREAM_INPUT_PATH}]]" in text
-                ]
+                source_links = [rel for rel, text in target_texts.items() if f"- [[{DREAM_INPUT_PATH}]]" in text]
                 print(f"[wikilink] integrated targets: {target_paths}")
                 print(f"[wikilink] integrated targets with [[digest/...]] links: {digest_wikilinks}")
-                print(f"[wikilink] integrated targets with derived_from source links: {provenance_links}")
+                print(f"[wikilink] integrated targets with source links: {source_links}")
                 assert target_texts, f"no integrated target files found: {target_paths}"
-                assert provenance_links, (
-                    "no derived_from wikilink back to the changed daily note in integrated targets\n"
+                assert source_links, (
+                    "no source wikilink back to the changed daily note in integrated targets\n"
                     f"targets: {target_paths}"
                 )
                 assert digest_wikilinks, (
diff --git a/tests/unit/test_auto_dream.py b/tests/unit/test_auto_dream.py
index a6a15402..9577d999 100644
--- a/tests/unit/test_auto_dream.py
+++ b/tests/unit/test_auto_dream.py
@@ -11,7 +11,7 @@ from reme.components.application_context import ApplicationContext
 from reme.components.file_catalog import BaseFileCatalog
 from reme.components.file_store import BaseFileStore
 from reme.components.runtime_context import RuntimeContext
-from reme.schema import DreamState
+from reme.schema import DreamState, FileNode
 from reme.steps.evolve.dream.extract import DreamExtractStep
 from reme.steps.evolve.dream.finish import DreamFinishStep
 from reme.steps.evolve.dream.proactive import ProactiveStep
@@ -94,6 +94,48 @@ def test_scan_day_files_includes_nested_md_and_excludes_interests():
         ]
 
 
+def test_dream_extract_matches_posix_catalog_paths(tmp_path):
+    """Unchanged nested files retain their POSIX catalog entries on every platform."""
+
+    class Catalog(_Catalog):
+        """Catalog seeded with POSIX paths and recording deletions."""
+
+        def __init__(self, nodes):
+            super().__init__()
+            self.nodes = nodes
+            self.deleted = []
+
+        async def delete(self, path):
+            self.deleted.extend(path if isinstance(path, list) else [path])
+
+        async def get_nodes(self, paths=None):
+            return self.nodes
+
+    async def run():
+        note = _touch(tmp_path / "daily" / "2026-05-28" / "nested" / "session.md")
+        rel_path = note.relative_to(tmp_path).as_posix()
+        catalog = Catalog([FileNode(path=rel_path, st_mtime=note.stat().st_mtime)])
+        step = DreamExtractStep(scan_days=1, app_context=ApplicationContext(workspace_dir=str(tmp_path)))
+
+        with patch("reme.steps.evolve.dream.extract.refresh_day_index", return_value={}):
+            response = await step(
+                RuntimeContext(
+                    date="2026-05-28",
+                    file_catalog=catalog,
+                    file_store=_FileStore(tmp_path),
+                ),
+            )
+
+        dream = response.metadata["dream"]
+        assert response.success is True
+        assert dream["unchanged_paths"] == [rel_path]
+        assert dream["changed_paths"] == []
+        assert dream["deleted_paths"] == []
+        assert not catalog.deleted
+
+    asyncio.run(run())
+
+
 def test_recent_dates_includes_anchor_and_previous_days():
     """Recent date window is inclusive and chronological."""
     assert recent_dates("2026-05-28", 3) == ["2026-05-26", "2026-05-27", "2026-05-28"]
@@ -152,7 +194,7 @@ def test_extract_without_llm_marks_changed_paths_failed(tmp_path):
 
         dream = response.metadata["dream"]
         assert response.success is False
-        assert str(note.relative_to(tmp_path)) in dream["changed_paths"]
+        assert note.relative_to(tmp_path).as_posix() in dream["changed_paths"]
         assert dream["failed_paths"] == dream["changed_paths"]
 
     asyncio.run(run())
@@ -299,9 +341,9 @@ def test_finish_does_not_checkpoint_failed_changed_paths():
                 dates=["2026-05-26", "2026-05-27", "2026-05-28"],
                 workspace=str(workspace),
                 daily_dir="daily",
-                changed_paths=[str(ok.relative_to(workspace)), str(failed.relative_to(workspace))],
-                failed_paths=[str(failed.relative_to(workspace))],
-                interests_paths=[str(interests.relative_to(workspace))],
+                changed_paths=[ok.relative_to(workspace).as_posix(), failed.relative_to(workspace).as_posix()],
+                failed_paths=[failed.relative_to(workspace).as_posix()],
+                interests_paths=[interests.relative_to(workspace).as_posix()],
                 integrate_results=[
                     {
                         "action": "CREATE",
@@ -320,10 +362,10 @@ def test_finish_does_not_checkpoint_failed_changed_paths():
             assert "topics:" not in resp.answer
             assert "Changes:" in resp.answer
             assert "- [digest/procedure/example.md][CREATE]: Created a concise procedure node." in resp.answer
-            assert str(ok.relative_to(workspace)) in upserted
-            assert str(failed.relative_to(workspace)) not in upserted
-            assert str(interests.relative_to(workspace)) in upserted
-            assert str(day_index.relative_to(workspace)) in upserted
+            assert ok.relative_to(workspace).as_posix() in upserted
+            assert failed.relative_to(workspace).as_posix() not in upserted
+            assert interests.relative_to(workspace).as_posix() in upserted
+            assert day_index.relative_to(workspace).as_posix() in upserted
             assert catalog.dumps == 1
 
     asyncio.run(run())
diff --git a/tests/unit/test_base_component.py b/tests/unit/test_base_component.py
index 6b9d994b..27ac4dd8 100644
--- a/tests/unit/test_base_component.py
+++ b/tests/unit/test_base_component.py
@@ -306,7 +306,7 @@ def test_to_workspace_relative_inside_workspace():
             mixin = ComponentMixin()
             abs_path = mixin.workspace_path / "sub" / "file.md"
             rel = mixin.to_workspace_relative(abs_path)
-            assert rel == str(abs_path.relative_to(mixin.workspace_path))
+            assert rel == abs_path.relative_to(mixin.workspace_path).as_posix()
         finally:
             os.chdir(old_cwd)
 
diff --git a/tests/unit/test_common_steps.py b/tests/unit/test_common_steps.py
index d6c4a13f..600ee2ca 100644
--- a/tests/unit/test_common_steps.py
+++ b/tests/unit/test_common_steps.py
@@ -42,12 +42,14 @@ def _run(coro):
     asyncio.run(coro)
 
 
-def _node(path: str, links: list[tuple[str, str | None, str | None]] | None = None) -> FileNode:
-    """Build a FileNode with (target_path, target_anchor, predicate) outgoing edges."""
+def _node(path: str, links: list[tuple[str, str | None]] | None = None) -> FileNode:
+    """Build a FileNode with (target_path, target_anchor) outgoing edges."""
     return FileNode(
         path=path,
         st_mtime=1.0,
-        links=[FileLink(source_path=path, target_path=t, target_anchor=a, predicate=p) for t, a, p in (links or [])],
+        links=[
+            FileLink(source_path=path, target_path=target, target_anchor=anchor) for target, anchor in (links or [])
+        ],
     )
 
 
@@ -197,7 +199,7 @@ def test_traverse_forward_depth_1():
         with tempfile.TemporaryDirectory() as tmp, _temp_chdir(tmp):
             store = await _make_store(
                 [
-                    _node("a.md", [("b.md", None, None), ("c.md", "intro", "ref")]),
+                    _node("a.md", [("b.md", None), ("c.md", "intro")]),
                     _node("b.md"),
                     _node("c.md"),
                 ],
@@ -207,9 +209,8 @@ def test_traverse_forward_depth_1():
             results = _edges(step)
             paths = {r["path"] for r in results}
             assert paths == {"b.md", "c.md"}
-            # The 'ref' edge should report its predicate/anchor.
+            # Anchors remain part of traversal metadata.
             c_edge = next(r for r in results if r["path"] == "c.md")
-            assert c_edge["predicate"] == "ref"
             assert c_edge["anchor"] == "intro"
             assert c_edge["via"] == "a.md"
             assert c_edge["depth"] == 1
@@ -226,8 +227,8 @@ def test_traverse_backward_returns_inlinks():
         with tempfile.TemporaryDirectory() as tmp, _temp_chdir(tmp):
             store = await _make_store(
                 [
-                    _node("a.md", [("b.md", None, None)]),
-                    _node("c.md", [("b.md", None, None)]),
+                    _node("a.md", [("b.md", None)]),
+                    _node("c.md", [("b.md", None)]),
                     _node("b.md"),
                 ],
             )
@@ -248,8 +249,8 @@ def test_traverse_depth_2_expands():
         with tempfile.TemporaryDirectory() as tmp, _temp_chdir(tmp):
             store = await _make_store(
                 [
-                    _node("a.md", [("b.md", None, None)]),
-                    _node("b.md", [("c.md", None, None)]),
+                    _node("a.md", [("b.md", None)]),
+                    _node("b.md", [("c.md", None)]),
                     _node("c.md"),
                 ],
             )
@@ -311,8 +312,8 @@ def test_traverse_both_directions():
         with tempfile.TemporaryDirectory() as tmp, _temp_chdir(tmp):
             store = await _make_store(
                 [
-                    _node("upstream.md", [("center.md", None, None)]),
-                    _node("center.md", [("downstream.md", None, None)]),
+                    _node("upstream.md", [("center.md", None)]),
+                    _node("center.md", [("downstream.md", None)]),
                     _node("downstream.md"),
                 ],
             )
diff --git a/tests/unit/test_default_file_chunker.py b/tests/unit/test_default_file_chunker.py
index 1c300048..bf22d77e 100644
--- a/tests/unit/test_default_file_chunker.py
+++ b/tests/unit/test_default_file_chunker.py
@@ -105,7 +105,6 @@ def test_parse_links_bare():
     assert link.source_path == "src.md"
     assert link.target_path == "note"
     assert link.target_anchor is None
-    assert link.predicate is None
     print("✓ test_parse_links_bare passed")
 
 
@@ -115,7 +114,6 @@ def test_parse_links_with_anchor():
     assert len(links) == 1
     assert links[0].target_path == "note"
     assert links[0].target_anchor == "section A"
-    assert links[0].predicate is None
     print("✓ test_parse_links_with_anchor passed")
 
 
@@ -137,47 +135,18 @@ def test_parse_links_anchor_and_alias():
     print("✓ test_parse_links_anchor_and_alias passed")
 
 
-def test_parse_links_predicate_simple():
-    """Dataview inline: predicate:: [[target]]."""
-    links = WikilinkHandler.extract_links("author:: [[Alice]]", "src.md")
-    assert len(links) == 1
-    assert links[0].predicate == "author"
-    assert links[0].target_path == "Alice"
-    assert links[0].target_anchor is None
-    print("✓ test_parse_links_predicate_simple passed")
-
-
-def test_parse_links_predicate_bracketed():
-    """Dataview inline-bracket: [predicate:: [[target]]]."""
-    links = WikilinkHandler.extract_links("text [author:: [[Alice]]] more", "src.md")
-    assert len(links) == 1
-    assert links[0].predicate == "author"
-    assert links[0].target_path == "Alice"
-    print("✓ test_parse_links_predicate_bracketed passed")
-
-
-def test_parse_links_predicate_bracketed_with_anchor():
-    """[predicate:: [[target_path#target_anchor]]] — combined form."""
+def test_parse_links_ignores_legacy_relation_wrappers():
+    """Legacy relation text remains compatible as ordinary wikilinks."""
     links = WikilinkHandler.extract_links(
-        "[predicate:: [[target_path#target_anchor]]]",
+        "related:: [[a]]\n- related:: [[b]]\n[related:: [[c#section]]]",
         "src.md",
     )
-    assert len(links) == 1
-    link = links[0]
-    assert link.source_path == "src.md"
-    assert link.predicate == "predicate"
-    assert link.target_path == "target_path"
-    assert link.target_anchor == "target_anchor"
-    print("✓ test_parse_links_predicate_bracketed_with_anchor passed")
-
-
-def test_parse_links_predicate_sticks_to_first():
-    """Line-level predicate covers all wikilinks in its value portion."""
-    links = WikilinkHandler.extract_links("pred:: [[a]] and bare [[b]]", "src.md")
-    assert len(links) == 2
-    assert links[0].predicate == "pred" and links[0].target_path == "a"
-    assert links[1].predicate == "pred" and links[1].target_path == "b"
-    print("✓ test_parse_links_predicate_sticks_to_first passed")
+    assert [(link.target_path, link.target_anchor) for link in links] == [
+        ("a", None),
+        ("b", None),
+        ("c", "section"),
+    ]
+    assert all(link.predicate is None for link in links)
 
 
 def test_parse_links_multiple_on_one_line():
@@ -190,6 +159,30 @@ def test_parse_links_multiple_on_one_line():
     print("✓ test_parse_links_multiple_on_one_line passed")
 
 
+def test_parse_wikilink_line_ranges():
+    """Workspace paths and supported line-range forms create FileLink edges."""
+    links = WikilinkHandler.extract_links(
+        "[[daily/2026-06-20/session.md]] [[notes/example.md#L9]] "
+        + "[[notes/example.md#L9-L10]] [[notes/example.md#L9-L10,L15-L20]]",
+        "src.md",
+    )
+    assert [(link.target_path, link.target_anchor) for link in links] == [
+        ("daily/2026-06-20/session.md", None),
+        ("notes/example.md", "L9"),
+        ("notes/example.md", "L9-L10"),
+        ("notes/example.md", "L9-L10,L15-L20"),
+    ]
+
+
+def test_parse_markdown_links_are_ignored():
+    """Ordinary Markdown links do not create FileLink edges."""
+    links = WikilinkHandler.extract_links(
+        "[plain](../wiki/a.md) [section](../wiki/a.md#section) [ranges](../wiki/b.md#L9-L10,L15-L20)",
+        "daily/note.md",
+    )
+    assert not links
+
+
 def test_parse_links_no_match():
     """Strings without [[]] yield no links, even if '::' appears."""
     assert len(WikilinkHandler.extract_links("no link here :: foo", "src.md")) == 0
@@ -198,15 +191,6 @@ def test_parse_links_no_match():
     print("✓ test_parse_links_no_match passed")
 
 
-def test_parse_links_predicate_with_underscore_and_digits():
-    """Predicate identifier accepts letters, digits, underscore (no dash per Dataview spec)."""
-    links = WikilinkHandler.extract_links("see_also2:: [[target]]", "src.md")
-    assert len(links) == 1
-    assert links[0].predicate == "see_also2"
-    assert links[0].target_path == "target"
-    print("✓ test_parse_links_predicate_with_underscore_and_digits passed")
-
-
 def test_parse_links_in_file():
     """Integration: parse() populates FileNode.links from file content."""
 
@@ -227,11 +211,8 @@ def test_parse_links_in_file():
         try:
             chunker = DefaultFileChunker()
             file_node, _ = await chunker.chunk(temp_path)
-            triples = {(link.predicate, link.target_path, link.target_anchor) for link in file_node.links}
-            assert (None, "alpha", None) in triples
-            assert (None, "beta", "h2") in triples
-            assert ("author", "Alice", None) in triples
-            assert ("ref", "paper", "chapter 1") in triples
+            pairs = {(link.target_path, link.target_anchor) for link in file_node.links}
+            assert pairs == {("alpha", None), ("beta", "h2"), ("Alice", None), ("paper", "chapter 1")}
             assert all(link.source_path == file_node.path for link in file_node.links)
             print("✓ test_parse_links_in_file passed")
         finally:
@@ -384,13 +365,11 @@ if __name__ == "__main__":
     test_parse_links_with_anchor()
     test_parse_links_alias_dropped()
     test_parse_links_anchor_and_alias()
-    test_parse_links_predicate_simple()
-    test_parse_links_predicate_bracketed()
-    test_parse_links_predicate_bracketed_with_anchor()
-    test_parse_links_predicate_sticks_to_first()
+    test_parse_links_ignores_legacy_relation_wrappers()
     test_parse_links_multiple_on_one_line()
+    test_parse_wikilink_line_ranges()
+    test_parse_markdown_links_are_ignored()
     test_parse_links_no_match()
-    test_parse_links_predicate_with_underscore_and_digits()
     test_parse_links_in_file()
     test_parse_links_empty_when_no_content()
     test_chunk_does_not_split_wikilink_at_boundary()
diff --git a/tests/unit/test_file_graph.py b/tests/unit/test_file_graph.py
index 00d377c5..a125d3fe 100644
--- a/tests/unit/test_file_graph.py
+++ b/tests/unit/test_file_graph.py
@@ -38,6 +38,28 @@ def make_node(path: str, links: list[tuple[str, str | None]] | None = None) -> F
     )
 
 
+def test_file_node_loads_legacy_predicate_without_persisting_it():
+    """Legacy predicates are accepted for compatibility but omitted from new snapshots."""
+    node = FileNode.model_validate(
+        {
+            "path": "a.md",
+            "st_mtime": 1.0,
+            "links": [
+                {
+                    "source_path": "a.md",
+                    "target_path": "b.md",
+                    "target_anchor": "intro",
+                    "predicate": "related",
+                },
+            ],
+        },
+    )
+    assert node.links == [
+        FileLink(source_path="a.md", target_path="b.md", target_anchor="intro", predicate="related"),
+    ]
+    assert "predicate" not in node.model_dump_json()
+
+
 # Both backends should satisfy the same BaseFileGraph contract.
 BACKENDS = [LocalFileGraph, NxFileGraph]
 
diff --git a/tests/unit/test_line_anchor.py b/tests/unit/test_line_anchor.py
new file mode 100644
index 00000000..3e678af7
--- /dev/null
+++ b/tests/unit/test_line_anchor.py
@@ -0,0 +1,31 @@
+"""Tests for GitHub-style line anchor parsing."""
+
+import pytest
+
+from reme.utils import format_line_anchor, parse_line_anchor
+
+
+def test_parse_supported_line_anchors():
+    """Single lines, continuous ranges, and comma-separated ranges parse."""
+    assert parse_line_anchor("L9") == [(9, 9)]
+    assert parse_line_anchor("L9-L10") == [(9, 10)]
+    assert parse_line_anchor("L9-L10,L15-L20") == [(9, 10), (15, 20)]
+
+
+def test_parse_merges_overlapping_and_adjacent_ranges():
+    """Normalization merges ranges whose covered lines touch."""
+    ranges = parse_line_anchor("L9-L12,L11-L15,L16")
+    assert ranges == [(9, 16)]
+    assert format_line_anchor(ranges) == "L9-L16"
+
+
+def test_non_line_heading_is_unchanged():
+    """Ordinary heading anchors are outside this parser's contract."""
+    assert parse_line_anchor("Introduction") is None
+
+
+@pytest.mark.parametrize("anchor", ["L0", "L10-L9", "L9-Lx", "L9,"])
+def test_invalid_line_anchor_rejected(anchor):
+    """Line-looking anchors fail clearly when malformed or out of range."""
+    with pytest.raises(ValueError):
+        parse_line_anchor(anchor)
diff --git a/tests/unit/test_link_expansion.py b/tests/unit/test_link_expansion.py
index ebbd6417..57de6aec 100644
--- a/tests/unit/test_link_expansion.py
+++ b/tests/unit/test_link_expansion.py
@@ -113,7 +113,7 @@ def test_expand_links_returns_outlinks_and_inlinks_with_meta():
             assert len(a_out) == 1
             assert a_out[0]["path"] == "B.md"
             assert a_out[0]["meta"] == {"name": "B Doc", "description": "beta"}
-            assert a_out[0]["edges"] == [{"predicate": None, "anchor": None}]
+            assert a_out[0]["anchors"] == []
             assert result["A.md"]["inlinks"] == []
 
             b_in = result["B.md"]["inlinks"]
@@ -184,13 +184,13 @@ def test_render_expansion_lines_empty_input_yields_empty_list():
 
 
 def test_render_expansion_lines_outlinks_only():
-    """Single outlink with meta + plain edge renders as 3 lines."""
+    """Single outlink with metadata renders without a redundant edge line."""
     expansion = {
         "outlinks": [
             {
                 "path": "B.md",
                 "meta": {"name": "B", "description": "beta"},
-                "edges": [{"predicate": None, "anchor": None}],
+                "anchors": [],
             },
         ],
         "inlinks": [],
@@ -199,20 +199,19 @@ def test_render_expansion_lines_outlinks_only():
     assert lines == [
         "  outlinks (1):",
         '    → B.md  name="B"  description="beta"',
-        "        via plain",
     ]
     print("✓ test_render_expansion_lines_outlinks_only passed")
 
 
-def test_render_expansion_lines_inlinks_only_with_predicate_and_anchor():
-    """Inlink edge with predicate + anchor renders via descriptor."""
+def test_render_expansion_lines_inlinks_only_with_anchor():
+    """Inlink anchor renders via descriptor."""
     expansion = {
         "outlinks": [],
         "inlinks": [
             {
                 "path": "src.md",
                 "meta": {},
-                "edges": [{"predicate": "references", "anchor": "intro"}],
+                "anchors": ["intro"],
             },
         ],
     }
@@ -220,26 +219,26 @@ def test_render_expansion_lines_inlinks_only_with_predicate_and_anchor():
     assert lines == [
         "  inlinks (1):",
         "    ← src.md  (no meta)",
-        "        via predicate=references, anchor=#intro",
+        "        via anchor=#intro",
     ]
-    print("✓ test_render_expansion_lines_inlinks_only_with_predicate_and_anchor passed")
+    print("✓ test_render_expansion_lines_inlinks_only_with_anchor passed")
 
 
 def test_render_expansion_lines_both_directions_in_order():
     """outlinks block precedes inlinks block."""
     expansion = {
         "outlinks": [
-            {"path": "out.md", "meta": {"name": "Out"}, "edges": [{"predicate": None, "anchor": None}]},
+            {"path": "out.md", "meta": {"name": "Out"}, "anchors": []},
         ],
         "inlinks": [
-            {"path": "in.md", "meta": {"description": "incoming"}, "edges": [{"predicate": None, "anchor": None}]},
+            {"path": "in.md", "meta": {"description": "incoming"}, "anchors": []},
         ],
     }
     lines = render_expansion_lines(expansion)
     assert lines[0] == "  outlinks (1):"
-    assert lines[3] == "  inlinks (1):"
+    assert lines[2] == "  inlinks (1):"
     assert lines[1].lstrip().startswith("→")
-    assert lines[4].lstrip().startswith("←")
+    assert lines[3].lstrip().startswith("←")
     print("✓ test_render_expansion_lines_both_directions_in_order passed")
 
 
@@ -250,5 +249,5 @@ if __name__ == "__main__":
     test_expand_links_node_without_meta_returns_empty_meta_dict()
     test_render_expansion_lines_empty_input_yields_empty_list()
     test_render_expansion_lines_outlinks_only()
-    test_render_expansion_lines_inlinks_only_with_predicate_and_anchor()
+    test_render_expansion_lines_inlinks_only_with_anchor()
     test_render_expansion_lines_both_directions_in_order()
diff --git a/tests/unit/test_markdown_file_chunker.py b/tests/unit/test_markdown_file_chunker.py
index 1d7bdcb6..118443ca 100644
--- a/tests/unit/test_markdown_file_chunker.py
+++ b/tests/unit/test_markdown_file_chunker.py
@@ -284,9 +284,9 @@ def test_parse_links_literal_targets():
             path = _write_md(tmp, "note.md", body)
             chunker = MarkdownFileChunker()
             node, _ = await chunker.chunk(path)
-            triples = {(link.target_path, link.target_anchor, link.predicate) for link in node.links}
-            assert ("topics/Alice.md", None, None) in triples
-            assert ("topics/Bob.md", "sec", None) in triples
+            pairs = {(link.target_path, link.target_anchor) for link in node.links}
+            assert ("topics/Alice.md", None) in pairs
+            assert ("topics/Bob.md", "sec") in pairs
             # source_path always equals the node's own path
             for link in node.links:
                 assert link.source_path == node.path
@@ -317,8 +317,8 @@ def test_parse_links_short_and_no_ext_kept_literally():
     asyncio.run(run())
 
 
-def test_parse_links_predicate_inline_and_line():
-    """Both `pred:: [[X]]` (line-level) and `[pred:: [[X]]]` (inline) propagate predicate."""
+def test_parse_links_legacy_relation_text_is_ignored():
+    """Old relation wrappers do not prevent their inner wikilinks from parsing."""
 
     async def run():
         with tempfile.TemporaryDirectory() as tmp, temp_chdir(tmp):
@@ -326,16 +326,14 @@ def test_parse_links_predicate_inline_and_line():
             path = _write_md(tmp, "note.md", body)
             chunker = MarkdownFileChunker()
             node, _ = await chunker.chunk(path)
-            pairs = {(link.target_path, link.predicate) for link in node.links}
-            assert ("A.md", "extends") in pairs
-            assert ("B.md", "concerns") in pairs
-        print("✓ test_parse_links_predicate_inline_and_line passed")
+            assert {link.target_path for link in node.links} == {"A.md", "B.md"}
+        print("✓ test_parse_links_legacy_relation_text_is_ignored passed")
 
     asyncio.run(run())
 
 
 def test_parse_links_deduped():
-    """Repeated wikilinks with the same (target, predicate, anchor) emit one FileLink."""
+    """Repeated wikilinks with the same target and anchor emit one FileLink."""
 
     async def run():
         with tempfile.TemporaryDirectory() as tmp, temp_chdir(tmp):
@@ -616,7 +614,7 @@ if __name__ == "__main__":
     test_parse_chunk_ids_match_node_chunk_ids()
     test_parse_links_literal_targets()
     test_parse_links_short_and_no_ext_kept_literally()
-    test_parse_links_predicate_inline_and_line()
+    test_parse_links_legacy_relation_text_is_ignored()
     test_parse_links_deduped()
     test_parse_min_chunk_byte_size_clamped()
     test_parse_embed_toc_prefixes_chunk_text()
diff --git a/tests/unit/test_read_with_neighbors.py b/tests/unit/test_read_with_neighbors.py
index 77ac5abb..56b94f0c 100644
--- a/tests/unit/test_read_with_neighbors.py
+++ b/tests/unit/test_read_with_neighbors.py
@@ -141,6 +141,28 @@ def test_read_with_neighbors_injects_block_and_metadata():
     _run(run())
 
 
+def test_read_with_neighbors_uses_posix_nested_path():
+    """Nested graph keys use POSIX separators on every platform."""
+
+    async def run():
+        with tempfile.TemporaryDirectory() as tmp, temp_chdir(tmp):
+            store = await _store_with(
+                {
+                    "notes/A.md": {"body": "See [[notes/B.md]].", "name": "A Doc"},
+                    "notes/B.md": {"body": "End node.", "name": "B Doc"},
+                },
+            )
+            resp = await _read(store, step_kwargs={"with_neighbors": True}, path="notes/A.md")
+
+            expansion = resp.metadata.get("link_expansion")
+            assert expansion is not None
+            assert "notes/A.md" in expansion
+            assert expansion["notes/A.md"]["outlinks"][0]["path"] == "notes/B.md"
+            await store.close()
+
+    _run(run())
+
+
 # -- on + zero neighbors -----------------------------------------------------
 
 
diff --git a/tests/unit/test_wikilink_utils.py b/tests/unit/test_wikilink_utils.py
index 06eebbd1..7a69cfff 100644
--- a/tests/unit/test_wikilink_utils.py
+++ b/tests/unit/test_wikilink_utils.py
@@ -1,4 +1,4 @@
-"""Tests for the wikilink helpers in ``reme.utils.wikilink_handler``.
+"""Tests for wikilink helpers in ``reme.utils.wikilink_handler``.
 
 Two pure async helpers used by file_move / file_delete:
 
@@ -196,8 +196,8 @@ def test_retarget_image_marker_preserved():
     asyncio.run(run())
 
 
-def test_retarget_dataview_predicate_preserved():
-    """Line-level + inline-bracketed Dataview predicates pass through outside ``[[..]]``."""
+def test_retarget_preserves_surrounding_relation_text():
+    """Retargeting changes only the inner wikilink target."""
 
     async def run():
         with tempfile.TemporaryDirectory() as tmp, temp_chdir(tmp):
@@ -212,11 +212,33 @@ def test_retarget_dataview_predicate_preserved():
             assert "colleague:: [[people/Alice.md]]" in body
             assert "[负责:: [[people/Alice.md]]]" in body
             await store.close()
-        print("✓ test_retarget_dataview_predicate_preserved passed")
+        print("✓ test_retarget_preserves_surrounding_relation_text passed")
 
     asyncio.run(run())
 
 
+def test_markdown_links_are_ignored():
+    """Only double-bracket wikilinks create graph edges or get rewritten."""
+    markdown_links = (
+        "[label](../wiki/example.md) "
+        "[label](../wiki/example.md#section) "
+        "[label](../wiki/example.md#L9-L10,L15-L20)"
+    )
+    original = f"{markdown_links} [[wiki/example.md#L9]]"
+
+    rewritten, count = WikilinkHandler.scan_and_rewrite(
+        original,
+        old="wiki/example.md",
+        new="archive/example.md",
+    )
+
+    assert [(link.target_path, link.target_anchor) for link in WikilinkHandler.extract_links(original, "note.md")] == [
+        ("wiki/example.md", "L9"),
+    ]
+    assert count == 1
+    assert rewritten == f"{markdown_links} [[archive/example.md#L9]]"
+
+
 def test_retarget_multiple_files_aggregate_counts():
     """links_changed sums across files; by_file lists per-file counts."""
 
@@ -387,7 +409,8 @@ if __name__ == "__main__":
     test_retarget_alias_preserved()
     test_retarget_anchor_and_alias_together()
     test_retarget_image_marker_preserved()
-    test_retarget_dataview_predicate_preserved()
+    test_retarget_preserves_surrounding_relation_text()
+    test_markdown_links_are_ignored()
     test_retarget_multiple_files_aggregate_counts()
     test_retarget_dry_run_does_not_write()
     test_retarget_scope_limits_sweep()