diff --git a/checkpoint.json b/checkpoint.json
index 55a1df7b2..427def135 100644
--- a/checkpoint.json
+++ b/checkpoint.json
@@ -1,37 +1,44 @@
 {
-  "timestamp": "2026-03-20T01:03:14.135683Z",
-  "current_node": "preflight_lint",
+  "timestamp": "2026-03-20T01:06:12.946111Z",
+  "current_node": "implement",
   "completed_nodes": [
     "start",
     "toolchain",
     "preflight_compile",
-    "preflight_lint"
+    "preflight_lint",
+    "implement"
   ],
   "node_retries": {
     "preflight_lint": 1,
+    "implement": 1,
     "preflight_compile": 1,
     "start": 1,
     "toolchain": 1
   },
   "context_values": {
-    "internal.thread_id": "preflight_compile",
+    "internal.thread_id": "preflight_lint",
     "graph.rankdir": "LR",
-    "current.preamble": "Goal: # Plan: Add node-level model validation + missing catalog aliases\n\n## Context\n\nRunning `fabro run` with `model=\"gpt-54\"` on a workflow node fails at runtime with `LLM error: Not found on anthropic: model: gpt-54`. Two issues contribute:\n\n1. `fabro validate` doesn't warn about unknown model names on **nodes** (only validates stylesheet models via `StylesheetModelKnownRule`)\n2. The catalog is missing hyphenated aliases like `gpt-54` for `gpt-5.4` (only has `gpt54`)\n\n## Step 1: Add hyphenated aliases to catalog\n\n**File: `lib/crates/fabro-llm/src/catalog.json`**\n\n| Model ID | Current aliases | Add |\n|---|---|---|\n| `gpt-5.4` (line 161) | `[\"gpt54\"]` | `\"gpt-54\"` |\n| `gpt-5.4-pro` (line 178) | `[\"gpt54-pro\"]` | `\"gpt-54-pro\"` |\n| `gpt-5.4-mini` (line 194) | `[\"gpt54-mini\"]` | `\"gpt-54-mini\"` |\n\n**File: `lib/crates/fabro-llm/src/catalog.rs`** — add alias resolution tests:\n- `gpt_54_hyphenated_alias` → asserts `get_model_info(\"gpt-54\")` resolves to `gpt-5.4`\n- `gpt_54_pro_hyphenated_alias` → same for `gpt-54-pro`\n- `gpt_54_mini_hyphenated_alias` → same for `gpt-54-mini`\n\nUpdate insta snapshots (`cargo insta review`) for `gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`.\n\n## Step 2: Add `NodeModelKnownRule`\n\n**File: `lib/crates/fabro-validate/src/rules.rs`**\n\nAdd `NodeModelKnownRule` right after `StylesheetModelKnownRule` (after line 977). Mirrors the stylesheet rule but iterates nodes:\n\n- Iterate `graph.nodes.values()`\n- If `node.model()` is `Some` and `get_model_info()` returns `None` → emit `Severity::Warning`\n- If `node.provider()` is `Some` and `Provider::from_str()` fails → emit `Severity::Warning`\n- Set `node_id: Some(node.id.clone())` on each diagnostic\n\nRegister `Box::new(NodeModelKnownRule)` in `built_in_rules()` (line 33, after `StylesheetModelKnownRule`).\n\n**Tests** (following existing `stylesheet_model_known_rule_*` pattern):\n- `node_model_known_rule_valid_model` — known model, no warnings\n- `node_model_known_rule_unknown_model` — unknown model, 1 warning with model name and node ID\n- `node_model_known_rule_alias` — alias like `\"opus\"`, no warnings\n- `node_model_known_rule_unknown_provider` — bad provider, 1 warning\n- `node_model_known_rule_no_model_no_provider` — plain node, no warnings\n\n## Step 3: Fix the workflow\n\n**File: `fabro/workflows/implement/workflow.fabro` line 19**\n\nChange `model=\"gpt-54\"` to `model=\"gpt-54\"` (now valid after Step 1 adds the alias). No change needed — it will just work.\n\n## Verification\n\n```bash\ncargo test -p fabro-llm\ncargo insta review          # accept updated snapshots\ncargo test -p fabro-validate\ncargo clippy --workspace -- -D warnings\ncargo fmt --check --all\n```\n\nThen run `fabro validate fabro/workflows/implement/workflow.fabro` to confirm no warnings.\n\n\n## Completed stages\n- **toolchain**: success\n  - Script: `command -v cargo >/dev/null || { curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y && sudo ln -sf $HOME/.cargo/bin/* /usr/local/bin/; }; cargo --version 2>&1`\n  - Stdout:\n    ```\n    cargo 1.94.0 (85eff7c80 2026-01-15)\n    ```\n  - Stderr: (empty)\n- **preflight_compile**: success\n  - Script: `cargo check -q --workspace 2>&1`\n  - Stdout: (empty)\n  - Stderr: (empty)\n",
+    "current.preamble": "Goal: # Plan: Add node-level model validation + missing catalog aliases\n\n## Context\n\nRunning `fabro run` with `model=\"gpt-54\"` on a workflow node fails at runtime with `LLM error: Not found on anthropic: model: gpt-54`. Two issues contribute:\n\n1. `fabro validate` doesn't warn about unknown model names on **nodes** (only validates stylesheet models via `StylesheetModelKnownRule`)\n2. The catalog is missing hyphenated aliases like `gpt-54` for `gpt-5.4` (only has `gpt54`)\n\n## Step 1: Add hyphenated aliases to catalog\n\n**File: `lib/crates/fabro-llm/src/catalog.json`**\n\n| Model ID | Current aliases | Add |\n|---|---|---|\n| `gpt-5.4` (line 161) | `[\"gpt54\"]` | `\"gpt-54\"` |\n| `gpt-5.4-pro` (line 178) | `[\"gpt54-pro\"]` | `\"gpt-54-pro\"` |\n| `gpt-5.4-mini` (line 194) | `[\"gpt54-mini\"]` | `\"gpt-54-mini\"` |\n\n**File: `lib/crates/fabro-llm/src/catalog.rs`** — add alias resolution tests:\n- `gpt_54_hyphenated_alias` → asserts `get_model_info(\"gpt-54\")` resolves to `gpt-5.4`\n- `gpt_54_pro_hyphenated_alias` → same for `gpt-54-pro`\n- `gpt_54_mini_hyphenated_alias` → same for `gpt-54-mini`\n\nUpdate insta snapshots (`cargo insta review`) for `gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`.\n\n## Step 2: Add `NodeModelKnownRule`\n\n**File: `lib/crates/fabro-validate/src/rules.rs`**\n\nAdd `NodeModelKnownRule` right after `StylesheetModelKnownRule` (after line 977). Mirrors the stylesheet rule but iterates nodes:\n\n- Iterate `graph.nodes.values()`\n- If `node.model()` is `Some` and `get_model_info()` returns `None` → emit `Severity::Warning`\n- If `node.provider()` is `Some` and `Provider::from_str()` fails → emit `Severity::Warning`\n- Set `node_id: Some(node.id.clone())` on each diagnostic\n\nRegister `Box::new(NodeModelKnownRule)` in `built_in_rules()` (line 33, after `StylesheetModelKnownRule`).\n\n**Tests** (following existing `stylesheet_model_known_rule_*` pattern):\n- `node_model_known_rule_valid_model` — known model, no warnings\n- `node_model_known_rule_unknown_model` — unknown model, 1 warning with model name and node ID\n- `node_model_known_rule_alias` — alias like `\"opus\"`, no warnings\n- `node_model_known_rule_unknown_provider` — bad provider, 1 warning\n- `node_model_known_rule_no_model_no_provider` — plain node, no warnings\n\n## Step 3: Fix the workflow\n\n**File: `fabro/workflows/implement/workflow.fabro` line 19**\n\nChange `model=\"gpt-54\"` to `model=\"gpt-54\"` (now valid after Step 1 adds the alias). No change needed — it will just work.\n\n## Verification\n\n```bash\ncargo test -p fabro-llm\ncargo insta review          # accept updated snapshots\ncargo test -p fabro-validate\ncargo clippy --workspace -- -D warnings\ncargo fmt --check --all\n```\n\nThen run `fabro validate fabro/workflows/implement/workflow.fabro` to confirm no warnings.\n\n\n## Completed stages\n- **toolchain**: success\n  - Script: `command -v cargo >/dev/null || { curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y && sudo ln -sf $HOME/.cargo/bin/* /usr/local/bin/; }; cargo --version 2>&1`\n  - Stdout:\n    ```\n    cargo 1.94.0 (85eff7c80 2026-01-15)\n    ```\n  - Stderr: (empty)\n- **preflight_compile**: success\n  - Script: `cargo check -q --workspace 2>&1`\n  - Stdout: (empty)\n  - Stderr: (empty)\n- **preflight_lint**: success\n  - Script: `cargo clippy -q --workspace -- -D warnings 2>&1`\n  - Stdout: (empty)\n  - Stderr: (empty)\n",
     "failure_signature": "",
     "thread.toolchain.current_node": "preflight_compile",
     "internal.retry_count.preflight_compile": 1,
     "thread.preflight_compile.current_node": "preflight_lint",
     "internal.fidelity": "compact",
+    "internal.retry_count.implement": 1,
     "graph.model_stylesheet": "\n            * { backend: api; model: claude-opus-4-6;}\n        ",
+    "last_stage": "implement",
+    "last_response": "Everything passes. Let me summarize the changes made:\n\n## Summary\n\n### Step 1: Added hyphenated aliases to catalog\n\n**`lib/crates/fabro-llm/src/catalog.json`** — Added three new aliases:\n- `\"gpt-54\"",
     "internal.node_visit_count": 1,
     "internal.retry_count.toolchain": 1,
+    "thread.preflight_lint.current_node": "implement",
+    "response.implement": "Everything passes. Let me summarize the changes made:\n\n## Summary\n\n### Step 1: Added hyphenated aliases to catalog\n\n**`lib/crates/fabro-llm/src/catalog.json`** — Added three new aliases:\n- `\"gpt-54\"` for `gpt-5.4` (alongside existing `\"gpt54\"`)\n- `\"gpt-54-pro\"` for `gpt-5.4-pro` (alongside existing `\"gpt54-pro\"`)\n- `\"gpt-54-mini\"` for `gpt-5.4-mini` (alongside existing `\"gpt54-mini\"`)\n\n**`lib/crates/fabro-llm/src/catalog.rs`** — Updated two insta snapshots (`gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`) to include the new aliases, and added three new alias resolution tests:\n- `gpt_54_hyphenated_alias` — asserts `get_model_info(\"gpt-54\")` resolves to `gpt-5.4`\n- `gpt_54_pro_hyphenated_alias` — asserts `get_model_info(\"gpt-54-pro\")` resolves to `gpt-5.4-pro`\n- `gpt_54_mini_hyphenated_alias` — asserts `get_model_info(\"gpt-54-mini\")` resolves to `gpt-5.4-mini`\n\n### Step 2: Added `NodeModelKnownRule`\n\n**`lib/crates/fabro-validate/src/rules.rs`** — Added a new `NodeModelKnownRule` lint rule that:\n- Iterates `graph.nodes.values()` \n- Warns when `node.model()` is `Some` and `get_model_info()` returns `None`\n- Warns when `node.provider()` is `Some` and `Provider::from_str()` fails\n- Sets `node_id: Some(node.id.clone())` on each diagnostic\n\nRegistered `Box::new(NodeModelKnownRule)` in `built_in_rules()` after `StylesheetModelKnownRule`.\n\nAdded five tests following the existing pattern:\n- `node_model_known_rule_valid_model` — known model → no warnings\n- `node_model_known_rule_unknown_model` — unknown model → 1 warning with model name and node ID\n- `node_model_known_rule_alias` — alias `\"opus\"` → no warnings\n- `node_model_known_rule_unknown_provider` — bad provider → 1 warning\n- `node_model_known_rule_no_model_no_provider` — plain node → no warnings\n\n### Step 3: Workflow fix\n\nNo change needed to `fabro/workflows/implement/workflow.fabro` — the `model=\"gpt-54\"` on line 19 is now valid after Step 1 added the alias.\n\n### Verification\n- `cargo test -p fabro-llm` — 41 tests pass ✅\n- `cargo test -p fabro-validate` — 147 tests pass ✅\n- `cargo clippy --workspace -- -D warnings` — clean ✅\n- `cargo fmt --check --all` — clean ✅",
     "internal.retry_count.start": 1,
     "command.stderr": "",
     "failure_class": "",
     "outcome": "success",
     "internal.retry_count.preflight_lint": 1,
     "command.output": "",
-    "current_node": "preflight_lint",
+    "current_node": "implement",
     "internal.run_id": "01KM4C6R7M1XZF086NV6F0S7JS",
     "thread.start.current_node": "toolchain",
     "graph.goal": "# Plan: Add node-level model validation + missing catalog aliases\n\n## Context\n\nRunning `fabro run` with `model=\"gpt-54\"` on a workflow node fails at runtime with `LLM error: Not found on anthropic: model: gpt-54`. Two issues contribute:\n\n1. `fabro validate` doesn't warn about unknown model names on **nodes** (only validates stylesheet models via `StylesheetModelKnownRule`)\n2. The catalog is missing hyphenated aliases like `gpt-54` for `gpt-5.4` (only has `gpt54`)\n\n## Step 1: Add hyphenated aliases to catalog\n\n**File: `lib/crates/fabro-llm/src/catalog.json`**\n\n| Model ID | Current aliases | Add |\n|---|---|---|\n| `gpt-5.4` (line 161) | `[\"gpt54\"]` | `\"gpt-54\"` |\n| `gpt-5.4-pro` (line 178) | `[\"gpt54-pro\"]` | `\"gpt-54-pro\"` |\n| `gpt-5.4-mini` (line 194) | `[\"gpt54-mini\"]` | `\"gpt-54-mini\"` |\n\n**File: `lib/crates/fabro-llm/src/catalog.rs`** — add alias resolution tests:\n- `gpt_54_hyphenated_alias` → asserts `get_model_info(\"gpt-54\")` resolves to `gpt-5.4`\n- `gpt_54_pro_hyphenated_alias` → same for `gpt-54-pro`\n- `gpt_54_mini_hyphenated_alias` → same for `gpt-54-mini`\n\nUpdate insta snapshots (`cargo insta review`) for `gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`.\n\n## Step 2: Add `NodeModelKnownRule`\n\n**File: `lib/crates/fabro-validate/src/rules.rs`**\n\nAdd `NodeModelKnownRule` right after `StylesheetModelKnownRule` (after line 977). Mirrors the stylesheet rule but iterates nodes:\n\n- Iterate `graph.nodes.values()`\n- If `node.model()` is `Some` and `get_model_info()` returns `None` → emit `Severity::Warning`\n- If `node.provider()` is `Some` and `Provider::from_str()` fails → emit `Severity::Warning`\n- Set `node_id: Some(node.id.clone())` on each diagnostic\n\nRegister `Box::new(NodeModelKnownRule)` in `built_in_rules()` (line 33, after `StylesheetModelKnownRule`).\n\n**Tests** (following existing `stylesheet_model_known_rule_*` pattern):\n- `node_model_known_rule_valid_model` — known model, no warnings\n- `node_model_known_rule_unknown_model` — unknown model, 1 warning with model name and node ID\n- `node_model_known_rule_alias` — alias like `\"opus\"`, no warnings\n- `node_model_known_rule_unknown_provider` — bad provider, 1 warning\n- `node_model_known_rule_no_model_no_provider` — plain node, no warnings\n\n## Step 3: Fix the workflow\n\n**File: `fabro/workflows/implement/workflow.fabro` line 19**\n\nChange `model=\"gpt-54\"` to `model=\"gpt-54\"` (now valid after Step 1 adds the alias). No change needed — it will just work.\n\n## Verification\n\n```bash\ncargo test -p fabro-llm\ncargo insta review          # accept updated snapshots\ncargo test -p fabro-validate\ncargo clippy --workspace -- -D warnings\ncargo fmt --check --all\n```\n\nThen run `fabro validate fabro/workflows/implement/workflow.fabro` to confirm no warnings.\n"
@@ -56,6 +63,30 @@
       "notes": "Script completed: cargo clippy -q --workspace -- -D warnings 2>&1",
       "duration_ms": 10405
     },
+    "implement": {
+      "status": "success",
+      "context_updates": {
+        "last_stage": "implement",
+        "response.implement": "Everything passes. Let me summarize the changes made:\n\n## Summary\n\n### Step 1: Added hyphenated aliases to catalog\n\n**`lib/crates/fabro-llm/src/catalog.json`** — Added three new aliases:\n- `\"gpt-54\"` for `gpt-5.4` (alongside existing `\"gpt54\"`)\n- `\"gpt-54-pro\"` for `gpt-5.4-pro` (alongside existing `\"gpt54-pro\"`)\n- `\"gpt-54-mini\"` for `gpt-5.4-mini` (alongside existing `\"gpt54-mini\"`)\n\n**`lib/crates/fabro-llm/src/catalog.rs`** — Updated two insta snapshots (`gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`) to include the new aliases, and added three new alias resolution tests:\n- `gpt_54_hyphenated_alias` — asserts `get_model_info(\"gpt-54\")` resolves to `gpt-5.4`\n- `gpt_54_pro_hyphenated_alias` — asserts `get_model_info(\"gpt-54-pro\")` resolves to `gpt-5.4-pro`\n- `gpt_54_mini_hyphenated_alias` — asserts `get_model_info(\"gpt-54-mini\")` resolves to `gpt-5.4-mini`\n\n### Step 2: Added `NodeModelKnownRule`\n\n**`lib/crates/fabro-validate/src/rules.rs`** — Added a new `NodeModelKnownRule` lint rule that:\n- Iterates `graph.nodes.values()` \n- Warns when `node.model()` is `Some` and `get_model_info()` returns `None`\n- Warns when `node.provider()` is `Some` and `Provider::from_str()` fails\n- Sets `node_id: Some(node.id.clone())` on each diagnostic\n\nRegistered `Box::new(NodeModelKnownRule)` in `built_in_rules()` after `StylesheetModelKnownRule`.\n\nAdded five tests following the existing pattern:\n- `node_model_known_rule_valid_model` — known model → no warnings\n- `node_model_known_rule_unknown_model` — unknown model → 1 warning with model name and node ID\n- `node_model_known_rule_alias` — alias `\"opus\"` → no warnings\n- `node_model_known_rule_unknown_provider` — bad provider → 1 warning\n- `node_model_known_rule_no_model_no_provider` — plain node → no warnings\n\n### Step 3: Workflow fix\n\nNo change needed to `fabro/workflows/implement/workflow.fabro` — the `model=\"gpt-54\"` on line 19 is now valid after Step 1 added the alias.\n\n### Verification\n- `cargo test -p fabro-llm` — 41 tests pass ✅\n- `cargo test -p fabro-validate` — 147 tests pass ✅\n- `cargo clippy --workspace -- -D warnings` — clean ✅\n- `cargo fmt --check --all` — clean ✅",
+        "last_response": "Everything passes. Let me summarize the changes made:\n\n## Summary\n\n### Step 1: Added hyphenated aliases to catalog\n\n**`lib/crates/fabro-llm/src/catalog.json`** — Added three new aliases:\n- `\"gpt-54\""
+      },
+      "notes": "Stage completed: implement",
+      "usage": {
+        "model": "claude-opus-4-6",
+        "input_tokens": 40563,
+        "output_tokens": 6968,
+        "cache_read_tokens": 698910,
+        "cache_write_tokens": 45227,
+        "reasoning_tokens": 67,
+        "cost": 1.1310449999999999
+      },
+      "files_touched": [
+        "/home/daytona/workspace/lib/crates/fabro-llm/src/catalog.json",
+        "/home/daytona/workspace/lib/crates/fabro-llm/src/catalog.rs",
+        "/home/daytona/workspace/lib/crates/fabro-validate/src/rules.rs"
+      ],
+      "duration_ms": 176731
+    },
     "start": {
       "status": "success",
       "duration_ms": 0
@@ -70,11 +101,12 @@
       "duration_ms": 39
     }
   },
-  "next_node_id": "implement",
+  "next_node_id": "simplify_opus",
   "node_visits": {
     "preflight_compile": 1,
     "toolchain": 1,
     "preflight_lint": 1,
-    "start": 1
+    "start": 1,
+    "implement": 1
   }
 }
\ No newline at end of file
diff --git a/nodes/implement/prompt.md b/nodes/implement/prompt.md
new file mode 100644
index 000000000..fcd0a2d38
--- /dev/null
+++ b/nodes/implement/prompt.md
@@ -0,0 +1,84 @@
+Goal: # Plan: Add node-level model validation + missing catalog aliases
+
+## Context
+
+Running `fabro run` with `model="gpt-54"` on a workflow node fails at runtime with `LLM error: Not found on anthropic: model: gpt-54`. Two issues contribute:
+
+1. `fabro validate` doesn't warn about unknown model names on **nodes** (only validates stylesheet models via `StylesheetModelKnownRule`)
+2. The catalog is missing hyphenated aliases like `gpt-54` for `gpt-5.4` (only has `gpt54`)
+
+## Step 1: Add hyphenated aliases to catalog
+
+**File: `lib/crates/fabro-llm/src/catalog.json`**
+
+| Model ID | Current aliases | Add |
+|---|---|---|
+| `gpt-5.4` (line 161) | `["gpt54"]` | `"gpt-54"` |
+| `gpt-5.4-pro` (line 178) | `["gpt54-pro"]` | `"gpt-54-pro"` |
+| `gpt-5.4-mini` (line 194) | `["gpt54-mini"]` | `"gpt-54-mini"` |
+
+**File: `lib/crates/fabro-llm/src/catalog.rs`** — add alias resolution tests:
+- `gpt_54_hyphenated_alias` → asserts `get_model_info("gpt-54")` resolves to `gpt-5.4`
+- `gpt_54_pro_hyphenated_alias` → same for `gpt-54-pro`
+- `gpt_54_mini_hyphenated_alias` → same for `gpt-54-mini`
+
+Update insta snapshots (`cargo insta review`) for `gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`.
+
+## Step 2: Add `NodeModelKnownRule`
+
+**File: `lib/crates/fabro-validate/src/rules.rs`**
+
+Add `NodeModelKnownRule` right after `StylesheetModelKnownRule` (after line 977). Mirrors the stylesheet rule but iterates nodes:
+
+- Iterate `graph.nodes.values()`
+- If `node.model()` is `Some` and `get_model_info()` returns `None` → emit `Severity::Warning`
+- If `node.provider()` is `Some` and `Provider::from_str()` fails → emit `Severity::Warning`
+- Set `node_id: Some(node.id.clone())` on each diagnostic
+
+Register `Box::new(NodeModelKnownRule)` in `built_in_rules()` (line 33, after `StylesheetModelKnownRule`).
+
+**Tests** (following existing `stylesheet_model_known_rule_*` pattern):
+- `node_model_known_rule_valid_model` — known model, no warnings
+- `node_model_known_rule_unknown_model` — unknown model, 1 warning with model name and node ID
+- `node_model_known_rule_alias` — alias like `"opus"`, no warnings
+- `node_model_known_rule_unknown_provider` — bad provider, 1 warning
+- `node_model_known_rule_no_model_no_provider` — plain node, no warnings
+
+## Step 3: Fix the workflow
+
+**File: `fabro/workflows/implement/workflow.fabro` line 19**
+
+Change `model="gpt-54"` to `model="gpt-54"` (now valid after Step 1 adds the alias). No change needed — it will just work.
+
+## Verification
+
+```bash
+cargo test -p fabro-llm
+cargo insta review          # accept updated snapshots
+cargo test -p fabro-validate
+cargo clippy --workspace -- -D warnings
+cargo fmt --check --all
+```
+
+Then run `fabro validate fabro/workflows/implement/workflow.fabro` to confirm no warnings.
+
+
+## Completed stages
+- **toolchain**: success
+  - Script: `command -v cargo >/dev/null || { curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y && sudo ln -sf $HOME/.cargo/bin/* /usr/local/bin/; }; cargo --version 2>&1`
+  - Stdout:
+    ```
+    cargo 1.94.0 (85eff7c80 2026-01-15)
+    ```
+  - Stderr: (empty)
+- **preflight_compile**: success
+  - Script: `cargo check -q --workspace 2>&1`
+  - Stdout: (empty)
+  - Stderr: (empty)
+- **preflight_lint**: success
+  - Script: `cargo clippy -q --workspace -- -D warnings 2>&1`
+  - Stdout: (empty)
+  - Stderr: (empty)
+
+
+Read the plan file referenced in the goal and implement every step. Make all the code changes described in the plan. Use red/green TDD.
\ No newline at end of file
diff --git a/nodes/implement/provider_used.json b/nodes/implement/provider_used.json
new file mode 100644
index 000000000..c1905ff10
--- /dev/null
+++ b/nodes/implement/provider_used.json
@@ -0,0 +1,5 @@
+{
+  "mode": "agent",
+  "provider": "anthropic",
+  "model": "claude-opus-4-6"
+}
\ No newline at end of file
diff --git a/nodes/implement/response.md b/nodes/implement/response.md
new file mode 100644
index 000000000..4f50b1c54
--- /dev/null
+++ b/nodes/implement/response.md
@@ -0,0 +1,42 @@
+Everything passes. Let me summarize the changes made:
+
+## Summary
+
+### Step 1: Added hyphenated aliases to catalog
+
+**`lib/crates/fabro-llm/src/catalog.json`** — Added three new aliases:
+- `"gpt-54"` for `gpt-5.4` (alongside existing `"gpt54"`)
+- `"gpt-54-pro"` for `gpt-5.4-pro` (alongside existing `"gpt54-pro"`)
+- `"gpt-54-mini"` for `gpt-5.4-mini` (alongside existing `"gpt54-mini"`)
+
+**`lib/crates/fabro-llm/src/catalog.rs`** — Updated two insta snapshots (`gpt_5_4_in_catalog` and `gpt_5_4_pro_in_catalog`) to include the new aliases, and added three new alias resolution tests:
+- `gpt_54_hyphenated_alias` — asserts `get_model_info("gpt-54")` resolves to `gpt-5.4`
+- `gpt_54_pro_hyphenated_alias` — asserts `get_model_info("gpt-54-pro")` resolves to `gpt-5.4-pro`
+- `gpt_54_mini_hyphenated_alias` — asserts `get_model_info("gpt-54-mini")` resolves to `gpt-5.4-mini`
+
+### Step 2: Added `NodeModelKnownRule`
+
+**`lib/crates/fabro-validate/src/rules.rs`** — Added a new `NodeModelKnownRule` lint rule that:
+- Iterates `graph.nodes.values()` 
+- Warns when `node.model()` is `Some` and `get_model_info()` returns `None`
+- Warns when `node.provider()` is `Some` and `Provider::from_str()` fails
+- Sets `node_id: Some(node.id.clone())` on each diagnostic
+
+Registered `Box::new(NodeModelKnownRule)` in `built_in_rules()` after `StylesheetModelKnownRule`.
+
+Added five tests following the existing pattern:
+- `node_model_known_rule_valid_model` — known model → no warnings
+- `node_model_known_rule_unknown_model` — unknown model → 1 warning with model name and node ID
+- `node_model_known_rule_alias` — alias `"opus"` → no warnings
+- `node_model_known_rule_unknown_provider` — bad provider → 1 warning
+- `node_model_known_rule_no_model_no_provider` — plain node → no warnings
+
+### Step 3: Workflow fix
+
+No change needed to `fabro/workflows/implement/workflow.fabro` — the `model="gpt-54"` on line 19 is now valid after Step 1 added the alias.
+
+### Verification
+- `cargo test -p fabro-llm` — 41 tests pass ✅
+- `cargo test -p fabro-validate` — 147 tests pass ✅
+- `cargo clippy --workspace -- -D warnings` — clean ✅
+- `cargo fmt --check --all` — clean ✅
\ No newline at end of file
diff --git a/nodes/implement/status.json b/nodes/implement/status.json
new file mode 100644
index 000000000..180a29f43
--- /dev/null
+++ b/nodes/implement/status.json
@@ -0,0 +1,6 @@
+{
+  "status": "success",
+  "notes": "Stage completed: implement",
+  "failure_reason": null,
+  "timestamp": "2026-03-20T01:06:12.945599+00:00"
+}
\ No newline at end of file