diff --git a/run.json b/run.json
index 62dcd1ab8..62560cbeb 100644
--- a/run.json
+++ b/run.json
@@ -492,7 +492,7 @@
     "kind": "running"
   },
   "status_updated_at": "2026-05-24T01:42:40.350880Z",
-  "last_event_at": "2026-05-24T02:25:22.127084Z",
+  "last_event_at": "2026-05-24T02:25:26.188632Z",
   "pending_control": null,
   "checkpoints": [
     {
@@ -1022,9 +1022,9 @@
       }
     },
     {
-      "seq": 0,
+      "seq": 2289,
       "checkpoint": {
-        "timestamp": "2026-05-24T02:25:22.174984Z",
+        "timestamp": "2026-05-24T02:25:26.184463Z",
         "current_node": "simplify_gpt",
         "completed_nodes": [
           "start",
@@ -1036,13 +1036,212 @@
           "simplify_gpt"
         ],
         "node_retries": {},
+        "context_values": {
+          "internal.retry_count.start": 0,
+          "internal.thread_id": "simplify_opus",
+          "current_node": "simplify_gpt",
+          "failure_signature": "",
+          "internal.run_id": "01KSBT48J14ZMK9HQN48SVMG3T",
+          "response.implement": "Implemented batch run archive/unarchive end-to-end.\n\nSummary:\n- Added OpenAPI endpoints:\n  - `POST /api/v1/runs/archive`\n  - `POST /api/v1/runs/unarchive`\n- Added batch lifecycle schemas and regenerated Rust/TypeScript API clients.\n- Implemented server batch handlers with:\n  - `RequiredUser` auth only\n  - pre-mutation validation for empty, >250, duplicate, and invalid IDs\n  - sequential fail-soft processing\n  - ordered per-run results and summary counts\n  - existing `operations::archive` / `operations::unarchive` behavior and events\n- Added server tests covering:\n  - archive/unarchive visibility changes\n  - mixed success/conflict/not_found results\n  - idempotent archive/unarchive outcomes\n  - invalid request rejection before mutation\n  - unauthenticated and worker-token rejection\n- Added web helpers:\n  - `archiveRuns(runIds)`\n  - `unarchiveRuns(runIds)`\n- Updated runs list bulk actions and board “Archive all” to use one batch request per action.\n- Preserved cache invalidation/toast behavior, with partial-success batch summaries.\n- Added frontend helper and route-summary tests.\n\nValidation run:\n- `cargo build -q -p fabro-api` ✅\n- `cargo check -q --workspace` ✅\n- `cargo +nightly-2026-04-14 fmt --check --all` ✅\n- `cargo +nightly-2026-04-14 clippy -q --workspace --all-targets -- -D warnings` ✅\n- `cargo nextest run -p fabro-api` ✅\n- Targeted `fabro-server` batch/single archive tests ✅\n- `cd lib/packages/fabro-api-client && bun run generate` ✅\n- `cd lib/packages/fabro-api-client && bun run typecheck` ✅\n- `cd apps/fabro-web && bun run typecheck` ✅\n- `cd apps/fabro-web && bun run test` ✅\n\nNote:\n- Full `cargo nextest run -p fabro-server` still fails on existing/unrelated `server::tests::get_graph_returns_svg`; it invokes the test binary as the graph render subprocess and gets `running 0 tests...` instead of SVG. The new batch lifecycle tests pass.",
+          "internal.retry_count.simplify_opus": 0,
+          "internal.work_dir": "/home/daytona/workspace/fabro",
+          "graph.model_stylesheet": "\n            * { model: claude-opus-4-7; }\n        ",
+          "failure_class": "",
+          "last_stage": "simplify_gpt",
+          "graph.goal": "---\ntitle: feat: Batch run archive actions\ntype: feat\nstatus: active\ndate: 2026-05-24\n---\n\n# feat: Batch Run Archive Actions\n\n## Overview\n\nAdd API support for archiving and unarchiving multiple runs in one request, then update the web list and board multi-run actions to use the new batch endpoints. The existing single-run archive and unarchive endpoints remain unchanged for run-scoped callers and direct lifecycle actions.\n\n## Problem Frame\n\nThe web UI currently performs multi-run archive/unarchive actions by issuing one lifecycle request per selected run. That works, but it puts batch orchestration in the browser, repeats request overhead, and leaves API/CLI/MCP consumers without a first-class batch contract. A bounded fail-soft batch endpoint gives the server ownership of the multi-run operation while preserving the independent event stream semantics of each run.\n\n## Requirements Trace\n\n- R1. Provide public API endpoints that archive and unarchive many runs in one request.\n- R2. Preserve existing single-run archive/unarchive behavior, including eligibility, idempotency, and emitted events.\n- R3. Return per-run results so mixed-success batches can be reported without rolling back successful items.\n- R4. Update web bulk actions to make one request per batch action instead of one request per run.\n- R5. Keep cache invalidation and toast behavior equivalent to the current UI.\n\n## Scope Boundaries\n\n- Do not make batch archive/unarchive transactional; each run remains an independent event stream.\n- Do not add new workflow event variants; successful batch items emit the same run archive/unarchive events as single-run actions.\n- Do not change the existing `POST /api/v1/runs/{id}/archive` or `POST /api/v1/runs/{id}/unarchive` contracts.\n- Do not add CLI commands in this change; this plan is limited to HTTP API and web UI usage.\n\n## Context & Research\n\n### Relevant Code and Patterns\n\n- OpenAPI is the source of truth for HTTP contracts in `docs/public/api-reference/fabro-api.yaml`.\n- Server lifecycle routes live in `lib/crates/fabro-server/src/server/handler/lifecycle.rs`; single-run archive/unarchive already funnel through `operations::archive` and `operations::unarchive`.\n- Batch routes that are not tied to one path run ID should use `RequiredUser`, not `RequireRunScopedOrRunTools`, because a run-scoped worker token cannot safely authorize mutation of arbitrary run IDs from a request body.\n- Web lifecycle helpers live in `apps/fabro-web/app/lib/run-actions.ts`; list and board multi-run archive behavior lives in `apps/fabro-web/app/routes/runs.tsx`.\n- Run-list cache invalidation already uses `mutateRunListCaches` in `apps/fabro-web/app/lib/board-cache.ts`.\n\n### Strategy Docs\n\n- `docs/internal/testing-strategy.md`: server tests should assert public HTTP contracts and prefer structured assertions.\n- `docs/internal/error-handling-strategy.md`: preserve structured errors internally and return curated API messages at HTTP boundaries.\n- `docs/internal/events-strategy.md`: no new events are needed because existing per-run lifecycle events already represent the durable state transition.\n\n## Key Technical Decisions\n\n- Add collection action endpoints `POST /api/v1/runs/archive` and `POST /api/v1/runs/unarchive`. These avoid changing single-run URLs and keep generated client methods clear (`batchArchiveRuns`, `batchUnarchiveRuns`).\n- Use fail-soft HTTP `200` responses for valid batch requests, even when individual items fail. Per-item failures carry structured result entries; request-level validation failures still return normal `400` errors.\n- Validate `run_ids` at the request boundary: non-empty, maximum 250 IDs, no duplicates, and every value parseable as a `RunId`. Invalid request bodies must not mutate any runs.\n- Process eligible IDs sequentially in server code. The batch is bounded, lifecycle operations append events, and sequential processing avoids adding lock-order or concurrency behavior that the feature does not need.\n- Treat idempotent single-run outcomes as successful batch outcomes: already archived counts as success for archive; terminal not-archived counts as success for unarchive.\n\n## API Contract\n\nAdd these OpenAPI operations:\n\n- `POST /api/v1/runs/archive`\n  - operationId: `batchArchiveRuns`\n  - request: `BatchRunLifecycleRequest`\n  - response: `BatchRunLifecycleResponse`\n- `POST /api/v1/runs/unarchive`\n  - operationId: `batchUnarchiveRuns`\n  - request: `BatchRunLifecycleRequest`\n  - response: `BatchRunLifecycleResponse`\n\nAdd schemas:\n\n- `BatchRunLifecycleRequest`\n  - required `run_ids`\n  - `run_ids`: array of strings, `minItems: 1`, `maxItems: 250`, `uniqueItems: true`\n- `BatchRunLifecycleResponse`\n  - required `results`, `summary`\n  - `results`: array of `BatchRunLifecycleResult`, ordered exactly like the request `run_ids`\n  - `summary`: `BatchRunLifecycleSummary`\n- `BatchRunLifecycleResult`\n  - required `run_id`, `ok`, `outcome`\n  - `run_id`: string\n  - `ok`: boolean\n  - `outcome`: enum covering `archived`, `already_archived`, `unarchived`, `not_archived`, `not_found`, `conflict`, `error`\n  - optional `run`: `Run`, present for successful items when a decorated summary can be loaded\n  - optional `error`: `ErrorResponseEntry`, present for failed items\n- `BatchRunLifecycleSummary`\n  - required `requested`, `succeeded`, `failed`\n  - all integer counts\n\n## Implementation Units\n\n- [ ] **Unit 1: OpenAPI batch lifecycle contract**\n\n**Goal:** Add the public API contract and generated clients for batch archive/unarchive.\n\n**Requirements:** R1, R3\n\n**Dependencies:** None\n\n**Files:**\n- Modify: `docs/public/api-reference/fabro-api.yaml`\n- Generated by build/codegen: `lib/crates/fabro-api/src/generated.rs`\n- Generated by codegen: `lib/packages/fabro-api-client/src/api/runs-api.ts`\n- Generated by codegen: `lib/packages/fabro-api-client/src/models/*`\n\n**Approach:**\n- Add the two collection action paths and the four batch lifecycle schemas described in the API Contract section.\n- Keep response status simple: `200` for a valid processed batch; `400`, `401`, and `500` for request-level failures.\n- Run Rust generation through `cargo build -p fabro-api`.\n- Run TypeScript generation through `cd lib/packages/fabro-api-client && bun run generate`.\n\n**Patterns to follow:**\n- Existing single-run archive/unarchive path docs in the same OpenAPI file.\n- Existing generated client workflow described in `AGENTS.md`.\n\n**Test scenarios:**\n- Happy path: generated Rust and TypeScript clients expose `batchArchiveRuns` and `batchUnarchiveRuns`.\n- Contract: request schema enforces `run_ids` as the only required input.\n- Contract: result entries can represent both a successful decorated `Run` and a failed `ErrorResponseEntry`.\n\n**Verification:**\n- API generation completes without hand-edited generated files.\n\n- [ ] **Unit 2: Server batch lifecycle handlers**\n\n**Goal:** Implement the two new endpoints in the server using existing archive/unarchive operations.\n\n**Requirements:** R1, R2, R3\n\n**Dependencies:** Unit 1\n\n**Files:**\n- Modify: `lib/crates/fabro-server/src/server/handler/lifecycle.rs`\n- Test: `lib/crates/fabro-server/src/server/tests.rs`\n\n**Approach:**\n- Register `POST /runs/archive` and `POST /runs/unarchive` in the lifecycle route set.\n- Use `RequiredUser` for both handlers and convert the user into `Principal::User` for per-run operation calls.\n- Add a small shared batch helper that accepts the parsed request, the action, and the actor, then returns `BatchRunLifecycleResponse`.\n- Before processing any item, validate the entire request for empty list, over-limit list, duplicate IDs, and invalid IDs. Return a normal `400` API error if validation fails.\n- For each valid ID, call `operations::archive` or `operations::unarchive`; map success outcomes to item-level success results and map `RunNotFound`/`Precondition` to item-level `not_found`/`conflict` failures.\n- For successful items, load and decorate the current run summary the same way single-run lifecycle responses do. If summary loading fails after the operation succeeds, record that item as `error` rather than hiding the failure.\n\n**Patterns to follow:**\n- `run_archive_action` for operation mapping and existing error semantics.\n- `run_response` / `state.decorate_run_summary` for response shape.\n- Existing server tests around `archive_and_unarchive_updates_listing_visibility`.\n\n**Test scenarios:**\n- Happy path: two terminal runs archived in one request return two successful result entries, summary `requested=2/succeeded=2/failed=0`, and both runs are hidden from default `GET /api/v1/runs`.\n- Happy path: two archived runs unarchived in one request return successful entries and both runs reappear in default listing.\n- Idempotency: archiving an already archived run returns `ok=true` with `already_archived`; unarchiving a terminal non-archived run returns `ok=true` with `not_archived`.\n- Mixed result: a batch containing one terminal run, one running run, and one missing run returns ordered results with one success, one `conflict`, and one `not_found`; the terminal run is still archived.\n- Error path: empty `run_ids`, duplicate IDs, invalid IDs, and more than 250 IDs return request-level `400` and mutate no runs.\n- Auth path: unauthenticated requests are rejected; run-scoped worker authentication is not accepted for batch endpoints.\n\n**Verification:**\n- Existing single-run archive/unarchive tests pass unchanged.\n- New endpoint tests prove both API contract and run-list visibility effects.\n\n- [ ] **Unit 3: Frontend lifecycle helpers**\n\n**Goal:** Add typed web helpers for batch archive/unarchive.\n\n**Requirements:** R3, R4, R5\n\n**Dependencies:** Unit 1\n\n**Files:**\n- Modify: `apps/fabro-web/app/lib/run-actions.ts`\n- Test: `apps/fabro-web/app/lib/run-actions.test.ts`\n\n**Approach:**\n- Add `archiveRuns(runIds: string[])` and `unarchiveRuns(runIds: string[])` wrappers around the generated client methods.\n- Keep existing `archiveRun` and `unarchiveRun` helpers unchanged.\n- Preserve existing `LifecycleActionError` behavior for single-run actions. Batch helpers should return the generated batch response for valid mixed results and throw only for request-level API failures.\n\n**Patterns to follow:**\n- Existing lifecycle action helpers in `run-actions.ts`.\n- Axios adapter tests in `run-actions.test.ts`.\n\n**Test scenarios:**\n- Happy path: `archiveRuns([\"run-1\", \"run-2\"])` sends one generated-client request and returns parsed batch results.\n- Mixed result: helper resolves a response containing one success and one per-item failure without throwing.\n- Request error: helper throws parsed `ApiError`/lifecycle-style data when the server returns a request-level `400`.\n- Regression: existing `archiveRun`, `unarchiveRun`, `canArchive`, and `canUnarchive` tests continue to pass.\n\n**Verification:**\n- Web unit tests cover the new helper contract without changing single-run behavior.\n\n- [ ] **Unit 4: Web bulk-action integration**\n\n**Goal:** Replace multi-request UI orchestration with one batch request per bulk action.\n\n**Requirements:** R4, R5\n\n**Dependencies:** Units 1 and 3\n\n**Files:**\n- Modify: `apps/fabro-web/app/routes/runs.tsx`\n- Test: `apps/fabro-web/app/routes/runs.test.tsx`\n\n**Approach:**\n- Update `BulkActionToolbar` to call `archiveRuns` or `unarchiveRuns` once with eligible selected IDs.\n- Add a small pure batch-summary helper in `runs.tsx`, export it for route tests, and use it to compute toast messages from the batch response summary rather than `Promise.allSettled`.\n- Keep current eligibility filtering: archive only selected runs whose lifecycle status is archivable, unarchive only selected archived runs.\n- Clear selection only when all eligible items succeed, matching the current all-success behavior.\n- Call `mutateRunListCaches` once after the batch settles.\n- Update the kanban column archive-all action to call `archiveRuns` once with the column’s eligible IDs and reuse the same success/partial/failure toast behavior where practical.\n\n**Patterns to follow:**\n- Current `BulkActionToolbar` pending state, clear-selection behavior, and toast wording.\n- Current `ColumnActionsMenu` archive-all action and cache invalidation.\n\n**Test scenarios:**\n- Happy path: selecting multiple archived rows and clicking Unarchive calls the batch helper once and clears selection after all items succeed.\n- Happy path: selecting multiple terminal rows and clicking Archive calls the batch helper once and invalidates run-list caches once.\n- Mixed result: partial failure keeps selection and shows an error-tone partial-success toast with succeeded/failed counts.\n- Ineligible selection: selecting only non-archivable runs still shows the existing “No selected runs can be archived” style error without making an API request.\n- Board action: archive-all for a column calls the batch helper once with all eligible IDs.\n\n**Verification:**\n- UI behavior remains equivalent from the user’s perspective, but network behavior becomes one request per bulk action.\n\n## System-Wide Impact\n\n- **Auth:** Batch endpoints are user-only. This intentionally avoids giving a worker token with one run scope the ability to mutate arbitrary run IDs from a request body.\n- **Events:** No new event names or payloads. Each successful item appends the existing per-run archive/unarchive event.\n- **Caching:** Frontend run-list caches are still invalidated after lifecycle changes. The batch path should reduce invalidation churn from once per selected run to once per user action.\n- **Generated clients:** Both Rust and TypeScript generated clients change because OpenAPI is the source of truth.\n- **Compatibility:** Single-run endpoints and existing generated methods remain available and unchanged.\n\n## Risks & Mitigations\n\n| Risk | Mitigation |\n|------|------------|\n| Route ambiguity with `/runs/{id}` paths | Use static collection action paths and add server tests that call the exact batch routes. |\n| Partial mutation surprises | Use explicit fail-soft result entries and summarize succeeded/failed counts. |\n| Worker token privilege expansion | Require `RequiredUser` for batch endpoints. |\n| Duplicate IDs producing confusing results | Reject duplicates at request validation before any mutation. |\n| UI toast regressions | Cover all-success, all-failure, partial-failure, and ineligible-selection behavior in frontend tests. |\n\n## Documentation / Operational Notes\n\n- The OpenAPI descriptions should explicitly state that batch actions are fail-soft and not transactional.\n- No migration, feature flag, or rollout sequencing is required.\n- No new public docs are required unless API reference publishing is part of the release process.\n\n## Sources & References\n\n- OpenAPI source: `docs/public/api-reference/fabro-api.yaml`\n- Server lifecycle handlers: `lib/crates/fabro-server/src/server/handler/lifecycle.rs`\n- Workflow archive operations: `lib/crates/fabro-workflow/src/operations/archive.rs`\n- Web lifecycle helpers: `apps/fabro-web/app/lib/run-actions.ts`\n- Web runs route: `apps/fabro-web/app/routes/runs.tsx`\n- Testing guidance: `docs/internal/testing-strategy.md`\n- Error handling guidance: `docs/internal/error-handling-strategy.md`\n- Event guidance: `docs/internal/events-strategy.md`\n",
+          "response.simplify_opus": "All checks pass. Here's a summary of what was fixed.\n\n## Summary\n\nLaunched three parallel review agents (reuse / quality / efficiency) and aggregated their findings. Applied the high-value fixes:\n\n### Rust (`lib/crates/fabro-server`)\n\n1. **Efficiency P0 — readiness resolved once per batch.** `batch_run_archive_item` previously called `state.decorate_run_summary(...)` inside the per-id loop, which re-resolves LLM credentials on every iteration (up to 250 LLM client builds per request). Now `batch_run_archive_action` calls `state.ask_fabro_readiness().await` once, then per-item uses `readiness.decorate(summary)` directly. Exposed `AskFabroReadiness`, `AskFabroReadiness::decorate`, and `AppState::ask_fabro_readiness` as `pub(crate)`.\n2. **Removed duplicate `WorkflowError → HTTP` mapping.** Extracted `archive_workflow_error_to_api_error(err) -> ApiError` and shared it between the new batch handler and the existing single-run `run_archive_action`.\n3. **Removed duplicate `operations::{archive,unarchive}` glue.** Extracted `run_archive_operation(state, id, actor, action)` returning `Result<BatchRunLifecycleResultOutcome, WorkflowError>` (the single-run handler discards the outcome). The single-run handler and batch handler now share this body.\n4. **Removed hand-rolled `batch_error_entry` builder.** Added `ApiError::into_response_entry() -> ErrorResponseEntry` in `error.rs` (one source of truth for the wire shape). Batch failures now construct standard `ApiError` instances (`ApiError::not_found`, `ApiError::new(StatusCode::CONFLICT, ...)`, etc.) and convert at the boundary. Dropped the `batch_success_result` / `batch_failure_result` builders; result entries are now inlined or built via `batch_result_failure(id, outcome, ApiError)`.\n5. Removed now-unused `Run` and `ErrorResponseEntry` imports from `lifecycle.rs`.\n\n### TypeScript (`apps/fabro-web`)\n\n6. **Removed local `runWord` helper**, switched all callers to the existing `plural(n, \"run\", \"runs\")` helper from `components/settings-panel`.\n7. **Simplified `summarizeBatchLifecycleAction`.** Dropped the redundant `total` parameter (was always either `summary.requested` or fabricated by callers) and the redundant `allSucceeded` wrapper (derivable from `summary.failed === 0`). Now takes a single `BatchRunLifecycleSummary` and returns the toast directly. Callers in `ColumnActionsMenu` and `BulkActionToolbar` updated; tests rewritten accordingly.\n\n### Skipped findings (intentional)\n\n- **`uniqueItems: true` on `BatchRunLifecycleRequest.run_ids`** — explicitly required by the implementation plan; the `as unknown as` cast workaround stays. The server-side dedup validation matches the schema constraint and produces a 400 with a clear message.\n- **Boilerplate handler pair `batch_archive_runs`/`batch_unarchive_runs`** — mirrors the existing single-run `archive_run`/`unarchive_run` pattern in the same file; collapsing would diverge from convention.\n- **`json_bearer_request` → `json_request` refactor** — pre-existing helper, low-priority; left alone.\n- **`BatchLifecycleLabel` stringly-typed verb** — localized to one helper, low impact; the simplification above already removed the redundant `.toLowerCase()` plumbing in callers.\n\n### Verification\n\n- `cargo check -p fabro-server --tests` — clean\n- `cargo +nightly-2026-04-14 clippy -p fabro-server --all-targets -- -D warnings` — clean\n- `cargo +nightly-2026-04-14 fmt -p fabro-server` — applied\n- `cargo nextest run -p fabro-server -E 'test(/archive|api_error/)'` — 11/11 pass (6 batch tests + 5 single-run/error tests)\n- `cd apps/fabro-web && bun test app/routes/runs.test.tsx app/lib/run-actions.test.ts` — 27/27 pass\n- `cd apps/fabro-web && bun run typecheck` — clean",
+          "thread.preflight_lint.current_node": "implement",
+          "internal.retry_count.toolchain": 0,
+          "last_response": "Reviewed the feature diff against `origin/main` for reuse, quality, and efficiency.\n\nResult: no additional cleanup changes were warranted; working tree remains clean.\n\nValidation passed:\n\n- `cargo nex",
+          "outcome": "succeeded",
+          "thread.preflight_compile.current_node": "preflight_lint",
+          "internal.retry_count.preflight_compile": 0,
+          "internal.fidelity": "compact",
+          "thread.implement.current_node": "simplify_opus",
+          "command.output": "blob://sha256/12ae32cb1ec02d01eda3581b127c1fee3b0dc53572ed6baf239721a03d82e126",
+          "internal.retry_count.implement": 0,
+          "thread.toolchain.current_node": "preflight_compile",
+          "thread.simplify_opus.current_node": "simplify_gpt",
+          "response.simplify_gpt": "Reviewed the feature diff against `origin/main` for reuse, quality, and efficiency.\n\nResult: no additional cleanup changes were warranted; working tree remains clean.\n\nValidation passed:\n\n- `cargo nextest run -p fabro-server ...`\n  - 6 focused lifecycle tests passed\n- `cd apps/fabro-web && bun test app/lib/run-actions.test.ts app/routes/runs.test.tsx`\n  - 27 frontend tests passed\n\nI also confirmed `git status --short` is clean after validation.",
+          "graph.rankdir": "LR",
+          "internal.retry_count.simplify_gpt": 0,
+          "thread.start.current_node": "toolchain",
+          "internal.retry_count.preflight_lint": 0,
+          "internal.node_visit_count": 1
+        },
+        "node_outcomes": {
+          "implement": {
+            "status": "succeeded",
+            "context_updates": {
+              "last_response": "Implemented batch run archive/unarchive end-to-end.\n\nSummary:\n- Added OpenAPI endpoints:\n  - `POST /api/v1/runs/archive`\n  - `POST /api/v1/runs/unarchive`\n- Added batch lifecycle schemas and regenerat",
+              "response.implement": "Implemented batch run archive/unarchive end-to-end.\n\nSummary:\n- Added OpenAPI endpoints:\n  - `POST /api/v1/runs/archive`\n  - `POST /api/v1/runs/unarchive`\n- Added batch lifecycle schemas and regenerated Rust/TypeScript API clients.\n- Implemented server batch handlers with:\n  - `RequiredUser` auth only\n  - pre-mutation validation for empty, >250, duplicate, and invalid IDs\n  - sequential fail-soft processing\n  - ordered per-run results and summary counts\n  - existing `operations::archive` / `operations::unarchive` behavior and events\n- Added server tests covering:\n  - archive/unarchive visibility changes\n  - mixed success/conflict/not_found results\n  - idempotent archive/unarchive outcomes\n  - invalid request rejection before mutation\n  - unauthenticated and worker-token rejection\n- Added web helpers:\n  - `archiveRuns(runIds)`\n  - `unarchiveRuns(runIds)`\n- Updated runs list bulk actions and board “Archive all” to use one batch request per action.\n- Preserved cache invalidation/toast behavior, with partial-success batch summaries.\n- Added frontend helper and route-summary tests.\n\nValidation run:\n- `cargo build -q -p fabro-api` ✅\n- `cargo check -q --workspace` ✅\n- `cargo +nightly-2026-04-14 fmt --check --all` ✅\n- `cargo +nightly-2026-04-14 clippy -q --workspace --all-targets -- -D warnings` ✅\n- `cargo nextest run -p fabro-api` ✅\n- Targeted `fabro-server` batch/single archive tests ✅\n- `cd lib/packages/fabro-api-client && bun run generate` ✅\n- `cd lib/packages/fabro-api-client && bun run typecheck` ✅\n- `cd apps/fabro-web && bun run typecheck` ✅\n- `cd apps/fabro-web && bun run test` ✅\n\nNote:\n- Full `cargo nextest run -p fabro-server` still fails on existing/unrelated `server::tests::get_graph_returns_svg`; it invokes the test binary as the graph render subprocess and gets `running 0 tests...` instead of SVG. The new batch lifecycle tests pass.",
+              "last_stage": "implement"
+            },
+            "notes": "Stage completed: implement",
+            "usage": {
+              "input": {
+                "usage": {
+                  "model": {
+                    "provider": "openai",
+                    "model_id": "gpt-5.5"
+                  },
+                  "tokens": {
+                    "input_tokens": 321093,
+                    "output_tokens": 20925,
+                    "reasoning_tokens": 13248,
+                    "cache_read_tokens": 24927232,
+                    "cache_write_tokens": 0
+                  }
+                },
+                "facts": {
+                  "algorithm": "openai"
+                }
+              },
+              "total_usd_micros": 15094271
+            }
+          },
+          "preflight_lint": {
+            "status": "succeeded",
+            "context_updates": {
+              "command.output": "blob://sha256/12ae32cb1ec02d01eda3581b127c1fee3b0dc53572ed6baf239721a03d82e126"
+            },
+            "notes": "Script completed: cargo +nightly-2026-04-14 clippy -q --workspace --all-targets -- -D warnings 2>&1",
+            "usage": null
+          },
+          "start": {
+            "status": "succeeded",
+            "usage": null
+          },
+          "toolchain": {
+            "status": "succeeded",
+            "context_updates": {
+              "command.output": "blob://sha256/fc14b2ba2d770e5cd3169df7a29525c962adfc4cfa3097b9098c63ebd61a748c"
+            },
+            "notes": "Script completed: 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",
+            "usage": null
+          },
+          "preflight_compile": {
+            "status": "succeeded",
+            "context_updates": {
+              "command.output": "blob://sha256/12ae32cb1ec02d01eda3581b127c1fee3b0dc53572ed6baf239721a03d82e126"
+            },
+            "notes": "Script completed: cargo check -q --workspace 2>&1",
+            "usage": null
+          },
+          "simplify_gpt": {
+            "status": "succeeded",
+            "context_updates": {
+              "last_response": "Reviewed the feature diff against `origin/main` for reuse, quality, and efficiency.\n\nResult: no additional cleanup changes were warranted; working tree remains clean.\n\nValidation passed:\n\n- `cargo nex",
+              "last_stage": "simplify_gpt",
+              "response.simplify_gpt": "Reviewed the feature diff against `origin/main` for reuse, quality, and efficiency.\n\nResult: no additional cleanup changes were warranted; working tree remains clean.\n\nValidation passed:\n\n- `cargo nextest run -p fabro-server ...`\n  - 6 focused lifecycle tests passed\n- `cd apps/fabro-web && bun test app/lib/run-actions.test.ts app/routes/runs.test.tsx`\n  - 27 frontend tests passed\n\nI also confirmed `git status --short` is clean after validation."
+            },
+            "notes": "Stage completed: simplify_gpt",
+            "usage": {
+              "input": {
+                "usage": {
+                  "model": {
+                    "provider": "openai",
+                    "model_id": "gpt-5.5"
+                  },
+                  "tokens": {
+                    "input_tokens": 148483,
+                    "output_tokens": 4449,
+                    "reasoning_tokens": 2599,
+                    "cache_read_tokens": 2595328,
+                    "cache_write_tokens": 0
+                  }
+                },
+                "facts": {
+                  "algorithm": "openai"
+                }
+              },
+              "total_usd_micros": 2251519
+            }
+          },
+          "simplify_opus": {
+            "status": "succeeded",
+            "context_updates": {
+              "last_stage": "simplify_opus",
+              "response.simplify_opus": "All checks pass. Here's a summary of what was fixed.\n\n## Summary\n\nLaunched three parallel review agents (reuse / quality / efficiency) and aggregated their findings. Applied the high-value fixes:\n\n### Rust (`lib/crates/fabro-server`)\n\n1. **Efficiency P0 — readiness resolved once per batch.** `batch_run_archive_item` previously called `state.decorate_run_summary(...)` inside the per-id loop, which re-resolves LLM credentials on every iteration (up to 250 LLM client builds per request). Now `batch_run_archive_action` calls `state.ask_fabro_readiness().await` once, then per-item uses `readiness.decorate(summary)` directly. Exposed `AskFabroReadiness`, `AskFabroReadiness::decorate`, and `AppState::ask_fabro_readiness` as `pub(crate)`.\n2. **Removed duplicate `WorkflowError → HTTP` mapping.** Extracted `archive_workflow_error_to_api_error(err) -> ApiError` and shared it between the new batch handler and the existing single-run `run_archive_action`.\n3. **Removed duplicate `operations::{archive,unarchive}` glue.** Extracted `run_archive_operation(state, id, actor, action)` returning `Result<BatchRunLifecycleResultOutcome, WorkflowError>` (the single-run handler discards the outcome). The single-run handler and batch handler now share this body.\n4. **Removed hand-rolled `batch_error_entry` builder.** Added `ApiError::into_response_entry() -> ErrorResponseEntry` in `error.rs` (one source of truth for the wire shape). Batch failures now construct standard `ApiError` instances (`ApiError::not_found`, `ApiError::new(StatusCode::CONFLICT, ...)`, etc.) and convert at the boundary. Dropped the `batch_success_result` / `batch_failure_result` builders; result entries are now inlined or built via `batch_result_failure(id, outcome, ApiError)`.\n5. Removed now-unused `Run` and `ErrorResponseEntry` imports from `lifecycle.rs`.\n\n### TypeScript (`apps/fabro-web`)\n\n6. **Removed local `runWord` helper**, switched all callers to the existing `plural(n, \"run\", \"runs\")` helper from `components/settings-panel`.\n7. **Simplified `summarizeBatchLifecycleAction`.** Dropped the redundant `total` parameter (was always either `summary.requested` or fabricated by callers) and the redundant `allSucceeded` wrapper (derivable from `summary.failed === 0`). Now takes a single `BatchRunLifecycleSummary` and returns the toast directly. Callers in `ColumnActionsMenu` and `BulkActionToolbar` updated; tests rewritten accordingly.\n\n### Skipped findings (intentional)\n\n- **`uniqueItems: true` on `BatchRunLifecycleRequest.run_ids`** — explicitly required by the implementation plan; the `as unknown as` cast workaround stays. The server-side dedup validation matches the schema constraint and produces a 400 with a clear message.\n- **Boilerplate handler pair `batch_archive_runs`/`batch_unarchive_runs`** — mirrors the existing single-run `archive_run`/`unarchive_run` pattern in the same file; collapsing would diverge from convention.\n- **`json_bearer_request` → `json_request` refactor** — pre-existing helper, low-priority; left alone.\n- **`BatchLifecycleLabel` stringly-typed verb** — localized to one helper, low impact; the simplification above already removed the redundant `.toLowerCase()` plumbing in callers.\n\n### Verification\n\n- `cargo check -p fabro-server --tests` — clean\n- `cargo +nightly-2026-04-14 clippy -p fabro-server --all-targets -- -D warnings` — clean\n- `cargo +nightly-2026-04-14 fmt -p fabro-server` — applied\n- `cargo nextest run -p fabro-server -E 'test(/archive|api_error/)'` — 11/11 pass (6 batch tests + 5 single-run/error tests)\n- `cd apps/fabro-web && bun test app/routes/runs.test.tsx app/lib/run-actions.test.ts` — 27/27 pass\n- `cd apps/fabro-web && bun run typecheck` — clean",
+              "last_response": "All checks pass. Here's a summary of what was fixed.\n\n## Summary\n\nLaunched three parallel review agents (reuse / quality / efficiency) and aggregated their findings. Applied the high-value fixes:\n\n###"
+            },
+            "notes": "Stage completed: simplify_opus",
+            "usage": {
+              "input": {
+                "usage": {
+                  "model": {
+                    "provider": "anthropic",
+                    "model_id": "claude-opus-4-7"
+                  },
+                  "tokens": {
+                    "input_tokens": 83032,
+                    "output_tokens": 25666,
+                    "reasoning_tokens": 0,
+                    "cache_read_tokens": 4039192,
+                    "cache_write_tokens": 497002
+                  }
+                },
+                "facts": {
+                  "algorithm": "anthropic",
+                  "cache_write_5m_tokens": 497002,
+                  "cache_write_1h_tokens": 0
+                }
+              },
+              "total_usd_micros": 6182668
+            },
+            "files_touched": [
+              "apps/fabro-web/app/routes/runs.test.tsx",
+              "apps/fabro-web/app/routes/runs.tsx",
+              "lib/crates/fabro-server/src/error.rs",
+              "lib/crates/fabro-server/src/server.rs",
+              "lib/crates/fabro-server/src/server/handler/lifecycle.rs"
+            ]
+          }
+        },
+        "next_node_id": "verify",
+        "git_commit_sha": "fec1c7d76c89bc334332b3e61a99fd240fe167c1",
+        "node_visits": {
+          "simplify_opus": 1,
+          "toolchain": 1,
+          "preflight_lint": 1,
+          "preflight_compile": 1,
+          "implement": 1,
+          "simplify_gpt": 1,
+          "start": 1
+        }
+      },
+      "diff": {
+        "summary": {
+          "files_changed": 16,
+          "additions": 1325,
+          "deletions": 87
+        }
+      }
+    },
+    {
+      "seq": 0,
+      "checkpoint": {
+        "timestamp": "2026-05-24T02:34:02.299678Z",
+        "current_node": "verify",
+        "completed_nodes": [
+          "start",
+          "toolchain",
+          "preflight_compile",
+          "preflight_lint",
+          "implement",
+          "simplify_opus",
+          "simplify_gpt",
+          "verify"
+        ],
+        "node_retries": {},
         "context_values": {
           "thread.preflight_lint.current_node": "implement",
           "thread.simplify_opus.current_node": "simplify_gpt",
           "thread.start.current_node": "toolchain",
           "failure_signature": "",
           "internal.retry_count.implement": 0,
-          "command.output": "blob://sha256/12ae32cb1ec02d01eda3581b127c1fee3b0dc53572ed6baf239721a03d82e126",
+          "command.output": "blob://sha256/0aa5b32b3afe68798ceba6562c5582733d8c5cc231f61f8d538d378b27be4127",
+          "internal.retry_count.verify": 0,
           "internal.retry_count.preflight_lint": 0,
           "response.implement": "Implemented batch run archive/unarchive end-to-end.\n\nSummary:\n- Added OpenAPI endpoints:\n  - `POST /api/v1/runs/archive`\n  - `POST /api/v1/runs/unarchive`\n- Added batch lifecycle schemas and regenerated Rust/TypeScript API clients.\n- Implemented server batch handlers with:\n  - `RequiredUser` auth only\n  - pre-mutation validation for empty, >250, duplicate, and invalid IDs\n  - sequential fail-soft processing\n  - ordered per-run results and summary counts\n  - existing `operations::archive` / `operations::unarchive` behavior and events\n- Added server tests covering:\n  - archive/unarchive visibility changes\n  - mixed success/conflict/not_found results\n  - idempotent archive/unarchive outcomes\n  - invalid request rejection before mutation\n  - unauthenticated and worker-token rejection\n- Added web helpers:\n  - `archiveRuns(runIds)`\n  - `unarchiveRuns(runIds)`\n- Updated runs list bulk actions and board “Archive all” to use one batch request per action.\n- Preserved cache invalidation/toast behavior, with partial-success batch summaries.\n- Added frontend helper and route-summary tests.\n\nValidation run:\n- `cargo build -q -p fabro-api` ✅\n- `cargo check -q --workspace` ✅\n- `cargo +nightly-2026-04-14 fmt --check --all` ✅\n- `cargo +nightly-2026-04-14 clippy -q --workspace --all-targets -- -D warnings` ✅\n- `cargo nextest run -p fabro-api` ✅\n- Targeted `fabro-server` batch/single archive tests ✅\n- `cd lib/packages/fabro-api-client && bun run generate` ✅\n- `cd lib/packages/fabro-api-client && bun run typecheck` ✅\n- `cd apps/fabro-web && bun run typecheck` ✅\n- `cd apps/fabro-web && bun run test` ✅\n\nNote:\n- Full `cargo nextest run -p fabro-server` still fails on existing/unrelated `server::tests::get_graph_returns_svg`; it invokes the test binary as the graph render subprocess and gets `running 0 tests...` instead of SVG. The new batch lifecycle tests pass.",
           "thread.implement.current_node": "simplify_opus",
@@ -1053,16 +1252,17 @@
           "thread.toolchain.current_node": "preflight_compile",
           "internal.retry_count.start": 0,
           "internal.run_id": "01KSBT48J14ZMK9HQN48SVMG3T",
-          "internal.thread_id": "simplify_opus",
+          "internal.thread_id": "simplify_gpt",
           "thread.preflight_compile.current_node": "preflight_lint",
           "internal.retry_count.simplify_gpt": 0,
           "last_stage": "simplify_gpt",
           "failure_class": "",
           "graph.model_stylesheet": "\n            * { model: claude-opus-4-7; }\n        ",
           "last_response": "Reviewed the feature diff against `origin/main` for reuse, quality, and efficiency.\n\nResult: no additional cleanup changes were warranted; working tree remains clean.\n\nValidation passed:\n\n- `cargo nex",
+          "thread.simplify_gpt.current_node": "verify",
           "graph.goal": "---\ntitle: feat: Batch run archive actions\ntype: feat\nstatus: active\ndate: 2026-05-24\n---\n\n# feat: Batch Run Archive Actions\n\n## Overview\n\nAdd API support for archiving and unarchiving multiple runs in one request, then update the web list and board multi-run actions to use the new batch endpoints. The existing single-run archive and unarchive endpoints remain unchanged for run-scoped callers and direct lifecycle actions.\n\n## Problem Frame\n\nThe web UI currently performs multi-run archive/unarchive actions by issuing one lifecycle request per selected run. That works, but it puts batch orchestration in the browser, repeats request overhead, and leaves API/CLI/MCP consumers without a first-class batch contract. A bounded fail-soft batch endpoint gives the server ownership of the multi-run operation while preserving the independent event stream semantics of each run.\n\n## Requirements Trace\n\n- R1. Provide public API endpoints that archive and unarchive many runs in one request.\n- R2. Preserve existing single-run archive/unarchive behavior, including eligibility, idempotency, and emitted events.\n- R3. Return per-run results so mixed-success batches can be reported without rolling back successful items.\n- R4. Update web bulk actions to make one request per batch action instead of one request per run.\n- R5. Keep cache invalidation and toast behavior equivalent to the current UI.\n\n## Scope Boundaries\n\n- Do not make batch archive/unarchive transactional; each run remains an independent event stream.\n- Do not add new workflow event variants; successful batch items emit the same run archive/unarchive events as single-run actions.\n- Do not change the existing `POST /api/v1/runs/{id}/archive` or `POST /api/v1/runs/{id}/unarchive` contracts.\n- Do not add CLI commands in this change; this plan is limited to HTTP API and web UI usage.\n\n## Context & Research\n\n### Relevant Code and Patterns\n\n- OpenAPI is the source of truth for HTTP contracts in `docs/public/api-reference/fabro-api.yaml`.\n- Server lifecycle routes live in `lib/crates/fabro-server/src/server/handler/lifecycle.rs`; single-run archive/unarchive already funnel through `operations::archive` and `operations::unarchive`.\n- Batch routes that are not tied to one path run ID should use `RequiredUser`, not `RequireRunScopedOrRunTools`, because a run-scoped worker token cannot safely authorize mutation of arbitrary run IDs from a request body.\n- Web lifecycle helpers live in `apps/fabro-web/app/lib/run-actions.ts`; list and board multi-run archive behavior lives in `apps/fabro-web/app/routes/runs.tsx`.\n- Run-list cache invalidation already uses `mutateRunListCaches` in `apps/fabro-web/app/lib/board-cache.ts`.\n\n### Strategy Docs\n\n- `docs/internal/testing-strategy.md`: server tests should assert public HTTP contracts and prefer structured assertions.\n- `docs/internal/error-handling-strategy.md`: preserve structured errors internally and return curated API messages at HTTP boundaries.\n- `docs/internal/events-strategy.md`: no new events are needed because existing per-run lifecycle events already represent the durable state transition.\n\n## Key Technical Decisions\n\n- Add collection action endpoints `POST /api/v1/runs/archive` and `POST /api/v1/runs/unarchive`. These avoid changing single-run URLs and keep generated client methods clear (`batchArchiveRuns`, `batchUnarchiveRuns`).\n- Use fail-soft HTTP `200` responses for valid batch requests, even when individual items fail. Per-item failures carry structured result entries; request-level validation failures still return normal `400` errors.\n- Validate `run_ids` at the request boundary: non-empty, maximum 250 IDs, no duplicates, and every value parseable as a `RunId`. Invalid request bodies must not mutate any runs.\n- Process eligible IDs sequentially in server code. The batch is bounded, lifecycle operations append events, and sequential processing avoids adding lock-order or concurrency behavior that the feature does not need.\n- Treat idempotent single-run outcomes as successful batch outcomes: already archived counts as success for archive; terminal not-archived counts as success for unarchive.\n\n## API Contract\n\nAdd these OpenAPI operations:\n\n- `POST /api/v1/runs/archive`\n  - operationId: `batchArchiveRuns`\n  - request: `BatchRunLifecycleRequest`\n  - response: `BatchRunLifecycleResponse`\n- `POST /api/v1/runs/unarchive`\n  - operationId: `batchUnarchiveRuns`\n  - request: `BatchRunLifecycleRequest`\n  - response: `BatchRunLifecycleResponse`\n\nAdd schemas:\n\n- `BatchRunLifecycleRequest`\n  - required `run_ids`\n  - `run_ids`: array of strings, `minItems: 1`, `maxItems: 250`, `uniqueItems: true`\n- `BatchRunLifecycleResponse`\n  - required `results`, `summary`\n  - `results`: array of `BatchRunLifecycleResult`, ordered exactly like the request `run_ids`\n  - `summary`: `BatchRunLifecycleSummary`\n- `BatchRunLifecycleResult`\n  - required `run_id`, `ok`, `outcome`\n  - `run_id`: string\n  - `ok`: boolean\n  - `outcome`: enum covering `archived`, `already_archived`, `unarchived`, `not_archived`, `not_found`, `conflict`, `error`\n  - optional `run`: `Run`, present for successful items when a decorated summary can be loaded\n  - optional `error`: `ErrorResponseEntry`, present for failed items\n- `BatchRunLifecycleSummary`\n  - required `requested`, `succeeded`, `failed`\n  - all integer counts\n\n## Implementation Units\n\n- [ ] **Unit 1: OpenAPI batch lifecycle contract**\n\n**Goal:** Add the public API contract and generated clients for batch archive/unarchive.\n\n**Requirements:** R1, R3\n\n**Dependencies:** None\n\n**Files:**\n- Modify: `docs/public/api-reference/fabro-api.yaml`\n- Generated by build/codegen: `lib/crates/fabro-api/src/generated.rs`\n- Generated by codegen: `lib/packages/fabro-api-client/src/api/runs-api.ts`\n- Generated by codegen: `lib/packages/fabro-api-client/src/models/*`\n\n**Approach:**\n- Add the two collection action paths and the four batch lifecycle schemas described in the API Contract section.\n- Keep response status simple: `200` for a valid processed batch; `400`, `401`, and `500` for request-level failures.\n- Run Rust generation through `cargo build -p fabro-api`.\n- Run TypeScript generation through `cd lib/packages/fabro-api-client && bun run generate`.\n\n**Patterns to follow:**\n- Existing single-run archive/unarchive path docs in the same OpenAPI file.\n- Existing generated client workflow described in `AGENTS.md`.\n\n**Test scenarios:**\n- Happy path: generated Rust and TypeScript clients expose `batchArchiveRuns` and `batchUnarchiveRuns`.\n- Contract: request schema enforces `run_ids` as the only required input.\n- Contract: result entries can represent both a successful decorated `Run` and a failed `ErrorResponseEntry`.\n\n**Verification:**\n- API generation completes without hand-edited generated files.\n\n- [ ] **Unit 2: Server batch lifecycle handlers**\n\n**Goal:** Implement the two new endpoints in the server using existing archive/unarchive operations.\n\n**Requirements:** R1, R2, R3\n\n**Dependencies:** Unit 1\n\n**Files:**\n- Modify: `lib/crates/fabro-server/src/server/handler/lifecycle.rs`\n- Test: `lib/crates/fabro-server/src/server/tests.rs`\n\n**Approach:**\n- Register `POST /runs/archive` and `POST /runs/unarchive` in the lifecycle route set.\n- Use `RequiredUser` for both handlers and convert the user into `Principal::User` for per-run operation calls.\n- Add a small shared batch helper that accepts the parsed request, the action, and the actor, then returns `BatchRunLifecycleResponse`.\n- Before processing any item, validate the entire request for empty list, over-limit list, duplicate IDs, and invalid IDs. Return a normal `400` API error if validation fails.\n- For each valid ID, call `operations::archive` or `operations::unarchive`; map success outcomes to item-level success results and map `RunNotFound`/`Precondition` to item-level `not_found`/`conflict` failures.\n- For successful items, load and decorate the current run summary the same way single-run lifecycle responses do. If summary loading fails after the operation succeeds, record that item as `error` rather than hiding the failure.\n\n**Patterns to follow:**\n- `run_archive_action` for operation mapping and existing error semantics.\n- `run_response` / `state.decorate_run_summary` for response shape.\n- Existing server tests around `archive_and_unarchive_updates_listing_visibility`.\n\n**Test scenarios:**\n- Happy path: two terminal runs archived in one request return two successful result entries, summary `requested=2/succeeded=2/failed=0`, and both runs are hidden from default `GET /api/v1/runs`.\n- Happy path: two archived runs unarchived in one request return successful entries and both runs reappear in default listing.\n- Idempotency: archiving an already archived run returns `ok=true` with `already_archived`; unarchiving a terminal non-archived run returns `ok=true` with `not_archived`.\n- Mixed result: a batch containing one terminal run, one running run, and one missing run returns ordered results with one success, one `conflict`, and one `not_found`; the terminal run is still archived.\n- Error path: empty `run_ids`, duplicate IDs, invalid IDs, and more than 250 IDs return request-level `400` and mutate no runs.\n- Auth path: unauthenticated requests are rejected; run-scoped worker authentication is not accepted for batch endpoints.\n\n**Verification:**\n- Existing single-run archive/unarchive tests pass unchanged.\n- New endpoint tests prove both API contract and run-list visibility effects.\n\n- [ ] **Unit 3: Frontend lifecycle helpers**\n\n**Goal:** Add typed web helpers for batch archive/unarchive.\n\n**Requirements:** R3, R4, R5\n\n**Dependencies:** Unit 1\n\n**Files:**\n- Modify: `apps/fabro-web/app/lib/run-actions.ts`\n- Test: `apps/fabro-web/app/lib/run-actions.test.ts`\n\n**Approach:**\n- Add `archiveRuns(runIds: string[])` and `unarchiveRuns(runIds: string[])` wrappers around the generated client methods.\n- Keep existing `archiveRun` and `unarchiveRun` helpers unchanged.\n- Preserve existing `LifecycleActionError` behavior for single-run actions. Batch helpers should return the generated batch response for valid mixed results and throw only for request-level API failures.\n\n**Patterns to follow:**\n- Existing lifecycle action helpers in `run-actions.ts`.\n- Axios adapter tests in `run-actions.test.ts`.\n\n**Test scenarios:**\n- Happy path: `archiveRuns([\"run-1\", \"run-2\"])` sends one generated-client request and returns parsed batch results.\n- Mixed result: helper resolves a response containing one success and one per-item failure without throwing.\n- Request error: helper throws parsed `ApiError`/lifecycle-style data when the server returns a request-level `400`.\n- Regression: existing `archiveRun`, `unarchiveRun`, `canArchive`, and `canUnarchive` tests continue to pass.\n\n**Verification:**\n- Web unit tests cover the new helper contract without changing single-run behavior.\n\n- [ ] **Unit 4: Web bulk-action integration**\n\n**Goal:** Replace multi-request UI orchestration with one batch request per bulk action.\n\n**Requirements:** R4, R5\n\n**Dependencies:** Units 1 and 3\n\n**Files:**\n- Modify: `apps/fabro-web/app/routes/runs.tsx`\n- Test: `apps/fabro-web/app/routes/runs.test.tsx`\n\n**Approach:**\n- Update `BulkActionToolbar` to call `archiveRuns` or `unarchiveRuns` once with eligible selected IDs.\n- Add a small pure batch-summary helper in `runs.tsx`, export it for route tests, and use it to compute toast messages from the batch response summary rather than `Promise.allSettled`.\n- Keep current eligibility filtering: archive only selected runs whose lifecycle status is archivable, unarchive only selected archived runs.\n- Clear selection only when all eligible items succeed, matching the current all-success behavior.\n- Call `mutateRunListCaches` once after the batch settles.\n- Update the kanban column archive-all action to call `archiveRuns` once with the column’s eligible IDs and reuse the same success/partial/failure toast behavior where practical.\n\n**Patterns to follow:**\n- Current `BulkActionToolbar` pending state, clear-selection behavior, and toast wording.\n- Current `ColumnActionsMenu` archive-all action and cache invalidation.\n\n**Test scenarios:**\n- Happy path: selecting multiple archived rows and clicking Unarchive calls the batch helper once and clears selection after all items succeed.\n- Happy path: selecting multiple terminal rows and clicking Archive calls the batch helper once and invalidates run-list caches once.\n- Mixed result: partial failure keeps selection and shows an error-tone partial-success toast with succeeded/failed counts.\n- Ineligible selection: selecting only non-archivable runs still shows the existing “No selected runs can be archived” style error without making an API request.\n- Board action: archive-all for a column calls the batch helper once with all eligible IDs.\n\n**Verification:**\n- UI behavior remains equivalent from the user’s perspective, but network behavior becomes one request per bulk action.\n\n## System-Wide Impact\n\n- **Auth:** Batch endpoints are user-only. This intentionally avoids giving a worker token with one run scope the ability to mutate arbitrary run IDs from a request body.\n- **Events:** No new event names or payloads. Each successful item appends the existing per-run archive/unarchive event.\n- **Caching:** Frontend run-list caches are still invalidated after lifecycle changes. The batch path should reduce invalidation churn from once per selected run to once per user action.\n- **Generated clients:** Both Rust and TypeScript generated clients change because OpenAPI is the source of truth.\n- **Compatibility:** Single-run endpoints and existing generated methods remain available and unchanged.\n\n## Risks & Mitigations\n\n| Risk | Mitigation |\n|------|------------|\n| Route ambiguity with `/runs/{id}` paths | Use static collection action paths and add server tests that call the exact batch routes. |\n| Partial mutation surprises | Use explicit fail-soft result entries and summarize succeeded/failed counts. |\n| Worker token privilege expansion | Require `RequiredUser` for batch endpoints. |\n| Duplicate IDs producing confusing results | Reject duplicates at request validation before any mutation. |\n| UI toast regressions | Cover all-success, all-failure, partial-failure, and ineligible-selection behavior in frontend tests. |\n\n## Documentation / Operational Notes\n\n- The OpenAPI descriptions should explicitly state that batch actions are fail-soft and not transactional.\n- No migration, feature flag, or rollout sequencing is required.\n- No new public docs are required unless API reference publishing is part of the release process.\n\n## Sources & References\n\n- OpenAPI source: `docs/public/api-reference/fabro-api.yaml`\n- Server lifecycle handlers: `lib/crates/fabro-server/src/server/handler/lifecycle.rs`\n- Workflow archive operations: `lib/crates/fabro-workflow/src/operations/archive.rs`\n- Web lifecycle helpers: `apps/fabro-web/app/lib/run-actions.ts`\n- Web runs route: `apps/fabro-web/app/routes/runs.tsx`\n- Testing guidance: `docs/internal/testing-strategy.md`\n- Error handling guidance: `docs/internal/error-handling-strategy.md`\n- Event guidance: `docs/internal/events-strategy.md`\n",
           "internal.retry_count.simplify_opus": 0,
-          "current_node": "simplify_gpt",
+          "current_node": "verify",
           "internal.retry_count.preflight_compile": 0,
           "internal.node_visit_count": 1,
           "graph.rankdir": "LR",
@@ -1086,6 +1286,14 @@
             "notes": "Script completed: cargo check -q --workspace 2>&1",
             "usage": null
           },
+          "verify": {
+            "status": "succeeded",
+            "context_updates": {
+              "command.output": "blob://sha256/0aa5b32b3afe68798ceba6562c5582733d8c5cc231f61f8d538d378b27be4127"
+            },
+            "notes": "Script completed: git fetch origin main 2>&1 && git merge --no-edit --no-stat origin/main 2>&1 && cargo +nightly-2026-04-14 fmt --all 2>&1 && cargo dev docs refresh 2>&1 && cargo +nightly-2026-04-14 fmt --check --all 2>&1 && { command -v rg >/dev/null 2>&1 || { echo 'rg is required for verify'; exit 127; }; } && ! rg -n 'AuthMode::Disabled|RunAuthMethod|RunSubjectProvenance|\\bActorRef\\b|\\bActorKind\\b|AuthenticatedSubject|AuthenticatedService|AuthorizeRunScoped|AuthorizeRunBlob|AuthorizeStageArtifact|AuthorizeCommandLog|auth_method\\s*==\\s*\"disabled\"' lib/crates apps lib/packages docs/public/api-reference/fabro-api.yaml 2>&1 && cargo +nightly-2026-04-14 clippy --workspace --all-targets -- -D warnings 2>&1 && cargo nextest run --workspace --status-level slow --profile ci 2>&1 && cargo dev docs check 2>&1 && bun install --frozen-lockfile 2>&1 && (cd apps/fabro-web && bun run typecheck) 2>&1 && (cd apps/fabro-web && bun run test) 2>&1 && (cd lib/packages/fabro-api-client && bun run typecheck) 2>&1 && cargo dev build -- -p fabro-cli --release 2>&1",
+            "usage": null
+          },
           "start": {
             "status": "succeeded",
             "usage": null
@@ -1198,15 +1406,16 @@
             }
           }
         },
-        "next_node_id": "verify",
+        "next_node_id": "exit",
         "node_visits": {
-          "toolchain": 1,
           "implement": 1,
           "preflight_lint": 1,
           "start": 1,
           "preflight_compile": 1,
+          "simplify_gpt": 1,
+          "toolchain": 1,
           "simplify_opus": 1,
-          "simplify_gpt": 1
+          "verify": 1
         }
       },
       "diff": {}
@@ -1232,6 +1441,195 @@
   "superseded_by": null,
   "pending_interviews": {},
   "stages": {
+    "simplify_opus@1": {
+      "first_event_seq": 954,
+      "prompt": null,
+      "response": null,
+      "completion": {
+        "outcome": "succeeded",
+        "notes": "Stage completed: simplify_opus",
+        "failure_reason": null,
+        "timestamp": "2026-05-24T02:21:20.248646Z"
+      },
+      "provider_used": {
+        "mode": "agent",
+        "provider": "anthropic",
+        "model": "claude-opus-4-7"
+      },
+      "diff": null,
+      "script_invocation": null,
+      "script_timing": null,
+      "parallel_results": null,
+      "output": null,
+      "started_at": "2026-05-24T02:10:52.799657Z",
+      "handler": "agent",
+      "timing": {
+        "wall_time_ms": 627448,
+        "inference_time_ms": 0,
+        "tool_time_ms": 0,
+        "active_time_ms": 0
+      },
+      "usage": {
+        "input_tokens": 83032,
+        "output_tokens": 25666,
+        "total_tokens": 4644892,
+        "reasoning_tokens": 0,
+        "cache_read_tokens": 4039192,
+        "cache_write_tokens": 497002,
+        "total_usd_micros": 6182668
+      },
+      "model": {
+        "provider": "anthropic",
+        "model_id": "claude-opus-4-7"
+      },
+      "todos": {
+        "kind": "anthropic_tasks",
+        "list_id": "anthropic_tasks:0303bca2-612e-4729-86b3-afc5ac218ce8",
+        "items": [
+          {
+            "id": "1",
+            "status": "completed",
+            "order": 0,
+            "subject": "Code reuse review",
+            "description": "Agent 1: find duplication of existing utilities in the batch lifecycle change</description>\n</invoke>"
+          },
+          {
+            "id": "2",
+            "status": "completed",
+            "order": 1,
+            "subject": "Code quality review",
+            "description": "Agent 2: review for hacky patterns, redundant state, copy-paste in the batch lifecycle change</description>\n</invoke>"
+          },
+          {
+            "id": "3",
+            "status": "completed",
+            "order": 2,
+            "subject": "Efficiency review",
+            "description": "Agent 3: review for unnecessary work, missed concurrency, hot-path bloat in batch lifecycle</description>\n</invoke>"
+          }
+        ]
+      },
+      "subagents": [
+        {
+          "agent_id": "39c4b473",
+          "depth": 1,
+          "task": "You are doing a CODE REUSE review for a feature that adds batch run archive/unarchive endpoints.\n\nRead the diff at /tmp/review.diff. Working directory: /home/daytona/workspace/fabro.\n\nFor each new function or block of new logic, search the codebase for existing utilities that could replace it. Use grep liberally:\n- Rust new code is in lib/crates/fabro-server/src/server/handler/lifecycle.rs (helpers `validate_batch_run_ids`, `batch_run_archive_item`, `load_decorated_run_after_lifecycle_action`, `batch_success_result`, `batch_failure_result`, `batch_error_entry`, plus the new route handlers).\n- The existing `run_archive_action` and `run_response` helpers are right above in the same file. Check whether the new batch handler unnecessarily reimplements logic that could reuse `run_response` and/or unify the error mapping with `run_archive_action`.\n- Check for an existing ErrorResponseEntry constructor / ApiError-to-entry helper before accepting the new `batch_error_entry` (grep ErrorResponseEntry, ApiError::new usages, ApiError -> ErrorResponseEntry conversions).\n- TypeScript new code is in apps/fabro-web/app/lib/run-actions.ts (`archiveRuns`, `unarchiveRuns`, `batchRunLifecycleAction`) and apps/fabro-web/app/routes/runs.tsx (`summarizeBatchLifecycleAction`, `runWord`). Check for an existing pluralize helper.\n\nDo NOT fix anything. Just report a concise list of concrete findings with file:line references and a suggested existing utility / function to reuse. If something looks clean, say so. Keep the report short.",
+          "status": {
+            "kind": "completed",
+            "success": true,
+            "turns_used": 31
+          }
+        },
+        {
+          "agent_id": "f0c1c4ac",
+          "depth": 1,
+          "task": "You are doing a CODE QUALITY review for a feature that adds batch run archive/unarchive endpoints.\n\nRead the diff at /tmp/review.diff. Working directory: /home/daytona/workspace/fabro.\n\nLook for these specific anti-patterns:\n1. Redundant state, derivable cached values, unnecessary indirections.\n2. Parameter sprawl (e.g. helper functions accepting more args than needed).\n3. Copy-paste with slight variation — especially:\n   - The two new handlers `batch_archive_runs` and `batch_unarchive_runs` in lib/crates/fabro-server/src/server/handler/lifecycle.rs are nearly identical; could they be unified or simplified?\n   - The new `load_decorated_run_after_lifecycle_action` is similar to the existing `run_response` helper at the top of the same file.\n   - The error mapping in `batch_run_archive_item` (Precondition -> Conflict, RunNotFound -> NotFound, _ -> Error) mirrors the existing `run_archive_action`.\n   - The two batch helpers in run-actions.ts have an unusual `as unknown as BatchRunLifecycleRequest` cast around `{ run_ids: ... }`. Is there a cleaner way given the generated client uses Set<string>?\n4. Leaky abstractions / boundary breaks.\n5. Stringly-typed code or raw strings where enums exist (e.g. action=\"archive\"|\"unarchive\" union; `BatchLifecycleLabel = \"Archive\" | \"Unarchive\"` and then `.toLowerCase()` of it).\n6. Toast summary helper in runs.tsx (`summarizeBatchLifecycleAction`) — is the shape (toast + allSucceeded boolean) sensible or overly complex?\n7. The new test helper `json_request` in tests.rs — is there an existing one?\n\nBe aggressive (greenfield). Report a concise prioritized list with file:line refs and concrete recommendations. Do NOT fix anything. Keep the report short.",
+          "status": {
+            "kind": "completed",
+            "success": true,
+            "turns_used": 20
+          }
+        },
+        {
+          "agent_id": "afa913f3",
+          "depth": 1,
+          "task": "You are doing an EFFICIENCY review for a feature that adds batch run archive/unarchive endpoints.\n\nRead the diff at /tmp/review.diff. Working directory: /home/daytona/workspace/fabro.\n\nFocus on:\n1. Sequential vs concurrent: in lib/crates/fabro-server/src/server/handler/lifecycle.rs `batch_run_archive_action` processes IDs sequentially (the spec says \"Process eligible IDs sequentially in server code\"). Confirm the loop avoids any per-iteration redundant work (e.g. cloning state, recomputing things). Note: the spec explicitly chose sequential — do not recommend parallelizing.\n2. Redundant work in `batch_run_archive_item`: it calls `operations::archive` then `state.store.get_cached_summary` then `state.decorate_run_summary`. Compare to single-run `run_archive_action` which only calls operations and then `run_response`. Is there extra work, e.g. is decorate_run_summary itself doing N+1 IO that compounds in the batch loop?\n3. Web: in apps/fabro-web/app/routes/runs.tsx — the batch action now triggers `mutateRunListCaches(mutate)` once per batch (good). Confirm there's no duplicate invalidation or extra request triggered for unchanged things.\n4. Memory: any unbounded data structures? The batch is capped at 250 so the Vec/HashSet are bounded — that's fine.\n5. Unnecessary existence checks (TOCTOU).\n6. Hot-path bloat.\n\nBriefly grep `state.decorate_run_summary` and `get_cached_summary` to understand what they do (cost-wise).\n\nReport a concise prioritized list of concrete findings with file:line refs. Do NOT fix anything. Keep the report short.",
+          "status": {
+            "kind": "completed",
+            "success": true,
+            "turns_used": 48
+          }
+        }
+      ],
+      "context_window": {
+        "provider": "anthropic",
+        "model": "claude-opus-4-7",
+        "context_window_tokens": 1000000,
+        "input_tokens": 95501,
+        "usage_percent": 9.5501,
+        "count_method": "response_usage_scaled_breakdown",
+        "staleness": "live",
+        "generated_at": "2026-05-24T02:21:20.135894Z",
+        "event_seq": 1680,
+        "breakdown": [
+          {
+            "category": "system_prompt",
+            "tokens": 2438,
+            "usage_percent": 0.2438
+          },
+          {
+            "category": "tools",
+            "tokens": 2755,
+            "usage_percent": 0.2755
+          },
+          {
+            "category": "memory",
+            "tokens": 5755,
+            "usage_percent": 0.5755
+          },
+          {
+            "category": "conversation",
+            "tokens": 84545,
+            "usage_percent": 8.4545
+          },
+          {
+            "category": "other",
+            "tokens": 8,
+            "usage_percent": 0.0008
+          }
+        ],
+        "warnings": []
+      },
+      "state": "succeeded"
+    },
+    "preflight_compile@1": {
+      "first_event_seq": 33,
+      "prompt": null,
+      "response": null,
+      "completion": {
+        "outcome": "succeeded",
+        "notes": "Script completed: cargo check -q --workspace 2>&1",
+        "failure_reason": null,
+        "timestamp": "2026-05-24T01:44:51.331786Z"
+      },
+      "provider_used": null,
+      "diff": null,
+      "script_invocation": {
+        "script": "cargo check -q --workspace 2>&1",
+        "command": "exec 2>&1\ncargo check -q --workspace 2>&1",
+        "language": "shell"
+      },
+      "script_timing": {
+        "output": "blob://sha256/12ae32cb1ec02d01eda3581b127c1fee3b0dc53572ed6baf239721a03d82e126",
+        "exit_code": 0,
+        "duration_ms": 124084,
+        "termination": "exited",
+        "output_bytes": 0,
+        "live_streaming": false
+      },
+      "parallel_results": null,
+      "output": null,
+      "output_bytes": 0,
+      "live_streaming": false,
+      "termination": "exited",
+      "started_at": "2026-05-24T01:42:47.237349Z",
+      "handler": "command",
+      "timing": {
+        "wall_time_ms": 124094,
+        "inference_time_ms": 0,
+        "tool_time_ms": 0,
+        "active_time_ms": 0
+      },
+      "usage": {
+        "input_tokens": 0,
+        "output_tokens": 0,
+        "total_tokens": 0,
+        "reasoning_tokens": 0,
+        "cache_read_tokens": 0,
+        "cache_write_tokens": 0
+      },
+      "state": "succeeded"
+    },
     "implement@1": {
       "first_event_seq": 53,
       "prompt": null,
@@ -1816,200 +2214,16 @@
       },
       "state": "succeeded"
     },
-    "toolchain@1": {
-      "first_event_seq": 23,
-      "prompt": null,
-      "response": null,
-      "completion": {
-        "outcome": "succeeded",
-        "notes": "Script completed: 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",
-        "failure_reason": null,
-        "timestamp": "2026-05-24T01:42:43.730391Z"
-      },
-      "provider_used": null,
-      "diff": null,
-      "script_invocation": {
-        "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",
-        "command": "exec 2>&1\ncommand -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",
-        "language": "shell"
-      },
-      "script_timing": {
-        "output": "blob://sha256/fc14b2ba2d770e5cd3169df7a29525c962adfc4cfa3097b9098c63ebd61a748c",
-        "exit_code": 0,
-        "duration_ms": 1433,
-        "termination": "exited",
-        "output_bytes": 36,
-        "live_streaming": true
-      },
-      "parallel_results": null,
-      "output": null,
-      "output_bytes": 36,
-      "live_streaming": true,
-      "termination": "exited",
-      "started_at": "2026-05-24T01:42:42.285677Z",
-      "handler": "command",
-      "timing": {
-        "wall_time_ms": 1444,
-        "inference_time_ms": 0,
-        "tool_time_ms": 0,
-        "active_time_ms": 0
-      },
-      "usage": {
-        "input_tokens": 0,
-        "output_tokens": 0,
-        "total_tokens": 0,
-        "reasoning_tokens": 0,
-        "cache_read_tokens": 0,
-        "cache_write_tokens": 0
-      },
-      "state": "succeeded"
-    },
-    "simplify_opus@1": {
-      "first_event_seq": 954,
-      "prompt": null,
-      "response": null,
-      "completion": {
-        "outcome": "succeeded",
-        "notes": "Stage completed: simplify_opus",
-        "failure_reason": null,
-        "timestamp": "2026-05-24T02:21:20.248646Z"
-      },
-      "provider_used": {
-        "mode": "agent",
-        "provider": "anthropic",
-        "model": "claude-opus-4-7"
-      },
-      "diff": null,
-      "script_invocation": null,
-      "script_timing": null,
-      "parallel_results": null,
-      "output": null,
-      "started_at": "2026-05-24T02:10:52.799657Z",
-      "handler": "agent",
-      "timing": {
-        "wall_time_ms": 627448,
-        "inference_time_ms": 0,
-        "tool_time_ms": 0,
-        "active_time_ms": 0
-      },
-      "usage": {
-        "input_tokens": 83032,
-        "output_tokens": 25666,
-        "total_tokens": 4644892,
-        "reasoning_tokens": 0,
-        "cache_read_tokens": 4039192,
-        "cache_write_tokens": 497002,
-        "total_usd_micros": 6182668
-      },
-      "model": {
-        "provider": "anthropic",
-        "model_id": "claude-opus-4-7"
-      },
-      "todos": {
-        "kind": "anthropic_tasks",
-        "list_id": "anthropic_tasks:0303bca2-612e-4729-86b3-afc5ac218ce8",
-        "items": [
-          {
-            "id": "1",
-            "status": "completed",
-            "order": 0,
-            "subject": "Code reuse review",
-            "description": "Agent 1: find duplication of existing utilities in the batch lifecycle change</description>\n</invoke>"
-          },
-          {
-            "id": "2",
-            "status": "completed",
-            "order": 1,
-            "subject": "Code quality review",
-            "description": "Agent 2: review for hacky patterns, redundant state, copy-paste in the batch lifecycle change</description>\n</invoke>"
-          },
-          {
-            "id": "3",
-            "status": "completed",
-            "order": 2,
-            "subject": "Efficiency review",
-            "description": "Agent 3: review for unnecessary work, missed concurrency, hot-path bloat in batch lifecycle</description>\n</invoke>"
-          }
-        ]
-      },
-      "subagents": [
-        {
-          "agent_id": "39c4b473",
-          "depth": 1,
-          "task": "You are doing a CODE REUSE review for a feature that adds batch run archive/unarchive endpoints.\n\nRead the diff at /tmp/review.diff. Working directory: /home/daytona/workspace/fabro.\n\nFor each new function or block of new logic, search the codebase for existing utilities that could replace it. Use grep liberally:\n- Rust new code is in lib/crates/fabro-server/src/server/handler/lifecycle.rs (helpers `validate_batch_run_ids`, `batch_run_archive_item`, `load_decorated_run_after_lifecycle_action`, `batch_success_result`, `batch_failure_result`, `batch_error_entry`, plus the new route handlers).\n- The existing `run_archive_action` and `run_response` helpers are right above in the same file. Check whether the new batch handler unnecessarily reimplements logic that could reuse `run_response` and/or unify the error mapping with `run_archive_action`.\n- Check for an existing ErrorResponseEntry constructor / ApiError-to-entry helper before accepting the new `batch_error_entry` (grep ErrorResponseEntry, ApiError::new usages, ApiError -> ErrorResponseEntry conversions).\n- TypeScript new code is in apps/fabro-web/app/lib/run-actions.ts (`archiveRuns`, `unarchiveRuns`, `batchRunLifecycleAction`) and apps/fabro-web/app/routes/runs.tsx (`summarizeBatchLifecycleAction`, `runWord`). Check for an existing pluralize helper.\n\nDo NOT fix anything. Just report a concise list of concrete findings with file:line references and a suggested existing utility / function to reuse. If something looks clean, say so. Keep the report short.",
-          "status": {
-            "kind": "completed",
-            "success": true,
-            "turns_used": 31
-          }
-        },
-        {
-          "agent_id": "f0c1c4ac",
-          "depth": 1,
-          "task": "You are doing a CODE QUALITY review for a feature that adds batch run archive/unarchive endpoints.\n\nRead the diff at /tmp/review.diff. Working directory: /home/daytona/workspace/fabro.\n\nLook for these specific anti-patterns:\n1. Redundant state, derivable cached values, unnecessary indirections.\n2. Parameter sprawl (e.g. helper functions accepting more args than needed).\n3. Copy-paste with slight variation — especially:\n   - The two new handlers `batch_archive_runs` and `batch_unarchive_runs` in lib/crates/fabro-server/src/server/handler/lifecycle.rs are nearly identical; could they be unified or simplified?\n   - The new `load_decorated_run_after_lifecycle_action` is similar to the existing `run_response` helper at the top of the same file.\n   - The error mapping in `batch_run_archive_item` (Precondition -> Conflict, RunNotFound -> NotFound, _ -> Error) mirrors the existing `run_archive_action`.\n   - The two batch helpers in run-actions.ts have an unusual `as unknown as BatchRunLifecycleRequest` cast around `{ run_ids: ... }`. Is there a cleaner way given the generated client uses Set<string>?\n4. Leaky abstractions / boundary breaks.\n5. Stringly-typed code or raw strings where enums exist (e.g. action=\"archive\"|\"unarchive\" union; `BatchLifecycleLabel = \"Archive\" | \"Unarchive\"` and then `.toLowerCase()` of it).\n6. Toast summary helper in runs.tsx (`summarizeBatchLifecycleAction`) — is the shape (toast + allSucceeded boolean) sensible or overly complex?\n7. The new test helper `json_request` in tests.rs — is there an existing one?\n\nBe aggressive (greenfield). Report a concise prioritized list with file:line refs and concrete recommendations. Do NOT fix anything. Keep the report short.",
-          "status": {
-            "kind": "completed",
-            "success": true,
-            "turns_used": 20
-          }
-        },
-        {
-          "agent_id": "afa913f3",
-          "depth": 1,
-          "task": "You are doing an EFFICIENCY review for a feature that adds batch run archive/unarchive endpoints.\n\nRead the diff at /tmp/review.diff. Working directory: /home/daytona/workspace/fabro.\n\nFocus on:\n1. Sequential vs concurrent: in lib/crates/fabro-server/src/server/handler/lifecycle.rs `batch_run_archive_action` processes IDs sequentially (the spec says \"Process eligible IDs sequentially in server code\"). Confirm the loop avoids any per-iteration redundant work (e.g. cloning state, recomputing things). Note: the spec explicitly chose sequential — do not recommend parallelizing.\n2. Redundant work in `batch_run_archive_item`: it calls `operations::archive` then `state.store.get_cached_summary` then `state.decorate_run_summary`. Compare to single-run `run_archive_action` which only calls operations and then `run_response`. Is there extra work, e.g. is decorate_run_summary itself doing N+1 IO that compounds in the batch loop?\n3. Web: in apps/fabro-web/app/routes/runs.tsx — the batch action now triggers `mutateRunListCaches(mutate)` once per batch (good). Confirm there's no duplicate invalidation or extra request triggered for unchanged things.\n4. Memory: any unbounded data structures? The batch is capped at 250 so the Vec/HashSet are bounded — that's fine.\n5. Unnecessary existence checks (TOCTOU).\n6. Hot-path bloat.\n\nBriefly grep `state.decorate_run_summary` and `get_cached_summary` to understand what they do (cost-wise).\n\nReport a concise prioritized list of concrete findings with file:line refs. Do NOT fix anything. Keep the report short.",
-          "status": {
-            "kind": "completed",
-            "success": true,
-            "turns_used": 48
-          }
-        }
-      ],
-      "context_window": {
-        "provider": "anthropic",
-        "model": "claude-opus-4-7",
-        "context_window_tokens": 1000000,
-        "input_tokens": 95501,
-        "usage_percent": 9.5501,
-        "count_method": "response_usage_scaled_breakdown",
-        "staleness": "live",
-        "generated_at": "2026-05-24T02:21:20.135894Z",
-        "event_seq": 1680,
-        "breakdown": [
-          {
-            "category": "system_prompt",
-            "tokens": 2438,
-            "usage_percent": 0.2438
-          },
-          {
-            "category": "tools",
-            "tokens": 2755,
-            "usage_percent": 0.2755
-          },
-          {
-            "category": "memory",
-            "tokens": 5755,
-            "usage_percent": 0.5755
-          },
-          {
-            "category": "conversation",
-            "tokens": 84545,
-            "usage_percent": 8.4545
-          },
-          {
-            "category": "other",
-            "tokens": 8,
-            "usage_percent": 0.0008
-          }
-        ],
-        "warnings": []
-      },
-      "state": "succeeded"
-    },
     "simplify_gpt@1": {
       "first_event_seq": 1692,
       "prompt": null,
       "response": null,
-      "completion": null,
+      "completion": {
+        "outcome": "succeeded",
+        "notes": "Stage completed: simplify_gpt",
+        "failure_reason": null,
+        "timestamp": "2026-05-24T02:25:22.173983Z"
+      },
       "provider_used": {
         "mode": "agent",
         "provider": "openai",
@@ -2022,17 +2236,24 @@
       "output": null,
       "started_at": "2026-05-24T02:21:24.113485Z",
       "handler": "agent",
+      "timing": {
+        "wall_time_ms": 238060,
+        "inference_time_ms": 0,
+        "tool_time_ms": 0,
+        "active_time_ms": 0
+      },
       "usage": {
-        "input_tokens": 302251,
-        "output_tokens": 6032,
-        "total_tokens": 3014664,
-        "reasoning_tokens": 3021,
-        "cache_read_tokens": 2703360,
-        "cache_write_tokens": 0
+        "input_tokens": 148483,
+        "output_tokens": 4449,
+        "total_tokens": 2750859,
+        "reasoning_tokens": 2599,
+        "cache_read_tokens": 2595328,
+        "cache_write_tokens": 0,
+        "total_usd_micros": 2251519
       },
       "model": {
         "provider": "openai",
-        "model_id": "gpt-5.5-2026-04-23"
+        "model_id": "gpt-5.5"
       },
       "todos": {
         "kind": "openai_plan",
@@ -2300,42 +2521,42 @@
           }
         ]
       },
-      "state": "running"
+      "state": "succeeded"
     },
-    "preflight_compile@1": {
-      "first_event_seq": 33,
+    "toolchain@1": {
+      "first_event_seq": 23,
       "prompt": null,
       "response": null,
       "completion": {
         "outcome": "succeeded",
-        "notes": "Script completed: cargo check -q --workspace 2>&1",
+        "notes": "Script completed: 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",
         "failure_reason": null,
-        "timestamp": "2026-05-24T01:44:51.331786Z"
+        "timestamp": "2026-05-24T01:42:43.730391Z"
       },
       "provider_used": null,
       "diff": null,
       "script_invocation": {
-        "script": "cargo check -q --workspace 2>&1",
-        "command": "exec 2>&1\ncargo check -q --workspace 2>&1",
+        "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",
+        "command": "exec 2>&1\ncommand -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",
         "language": "shell"
       },
       "script_timing": {
-        "output": "blob://sha256/12ae32cb1ec02d01eda3581b127c1fee3b0dc53572ed6baf239721a03d82e126",
+        "output": "blob://sha256/fc14b2ba2d770e5cd3169df7a29525c962adfc4cfa3097b9098c63ebd61a748c",
         "exit_code": 0,
-        "duration_ms": 124084,
+        "duration_ms": 1433,
         "termination": "exited",
-        "output_bytes": 0,
-        "live_streaming": false
+        "output_bytes": 36,
+        "live_streaming": true
       },
       "parallel_results": null,
       "output": null,
-      "output_bytes": 0,
-      "live_streaming": false,
+      "output_bytes": 36,
+      "live_streaming": true,
       "termination": "exited",
-      "started_at": "2026-05-24T01:42:47.237349Z",
+      "started_at": "2026-05-24T01:42:42.285677Z",
       "handler": "command",
       "timing": {
-        "wall_time_ms": 124094,
+        "wall_time_ms": 1444,
         "inference_time_ms": 0,
         "tool_time_ms": 0,
         "active_time_ms": 0
@@ -2398,6 +2619,33 @@
       },
       "state": "succeeded"
     },
+    "verify@1": {
+      "first_event_seq": 2292,
+      "prompt": null,
+      "response": null,
+      "completion": null,
+      "provider_used": null,
+      "diff": null,
+      "script_invocation": {
+        "script": "git fetch origin main 2>&1 && git merge --no-edit --no-stat origin/main 2>&1 && cargo +nightly-2026-04-14 fmt --all 2>&1 && cargo dev docs refresh 2>&1 && cargo +nightly-2026-04-14 fmt --check --all 2>&1 && { command -v rg >/dev/null 2>&1 || { echo 'rg is required for verify'; exit 127; }; } && ! rg -n 'AuthMode::Disabled|RunAuthMethod|RunSubjectProvenance|\\bActorRef\\b|\\bActorKind\\b|AuthenticatedSubject|AuthenticatedService|AuthorizeRunScoped|AuthorizeRunBlob|AuthorizeStageArtifact|AuthorizeCommandLog|auth_method\\s*==\\s*\"disabled\"' lib/crates apps lib/packages docs/public/api-reference/fabro-api.yaml 2>&1 && cargo +nightly-2026-04-14 clippy --workspace --all-targets -- -D warnings 2>&1 && cargo nextest run --workspace --status-level slow --profile ci 2>&1 && cargo dev docs check 2>&1 && bun install --frozen-lockfile 2>&1 && (cd apps/fabro-web && bun run typecheck) 2>&1 && (cd apps/fabro-web && bun run test) 2>&1 && (cd lib/packages/fabro-api-client && bun run typecheck) 2>&1 && cargo dev build -- -p fabro-cli --release 2>&1",
+        "command": "exec 2>&1\ngit fetch origin main 2>&1 && git merge --no-edit --no-stat origin/main 2>&1 && cargo +nightly-2026-04-14 fmt --all 2>&1 && cargo dev docs refresh 2>&1 && cargo +nightly-2026-04-14 fmt --check --all 2>&1 && { command -v rg >/dev/null 2>&1 || { echo 'rg is required for verify'; exit 127; }; } && ! rg -n 'AuthMode::Disabled|RunAuthMethod|RunSubjectProvenance|\\bActorRef\\b|\\bActorKind\\b|AuthenticatedSubject|AuthenticatedService|AuthorizeRunScoped|AuthorizeRunBlob|AuthorizeStageArtifact|AuthorizeCommandLog|auth_method\\s*==\\s*\"disabled\"' lib/crates apps lib/packages docs/public/api-reference/fabro-api.yaml 2>&1 && cargo +nightly-2026-04-14 clippy --workspace --all-targets -- -D warnings 2>&1 && cargo nextest run --workspace --status-level slow --profile ci 2>&1 && cargo dev docs check 2>&1 && bun install --frozen-lockfile 2>&1 && (cd apps/fabro-web && bun run typecheck) 2>&1 && (cd apps/fabro-web && bun run test) 2>&1 && (cd lib/packages/fabro-api-client && bun run typecheck) 2>&1 && cargo dev build -- -p fabro-cli --release 2>&1",
+        "language": "shell"
+      },
+      "script_timing": null,
+      "parallel_results": null,
+      "output": null,
+      "started_at": "2026-05-24T02:25:26.188178Z",
+      "handler": "command",
+      "usage": {
+        "input_tokens": 0,
+        "output_tokens": 0,
+        "total_tokens": 0,
+        "reasoning_tokens": 0,
+        "cache_read_tokens": 0,
+        "cache_write_tokens": 0
+      },
+      "state": "running"
+    },
     "start@1": {
       "first_event_seq": 19,
       "prompt": null,
diff --git a/stages/007-simplify_gpt@1/response.md b/stages/007-simplify_gpt@1/response.md
new file mode 100644
index 000000000..408aeb2c8
--- /dev/null
+++ b/stages/007-simplify_gpt@1/response.md
@@ -0,0 +1,12 @@
+Reviewed the feature diff against `origin/main` for reuse, quality, and efficiency.
+
+Result: no additional cleanup changes were warranted; working tree remains clean.
+
+Validation passed:
+
+- `cargo nextest run -p fabro-server ...`
+  - 6 focused lifecycle tests passed
+- `cd apps/fabro-web && bun test app/lib/run-actions.test.ts app/routes/runs.test.tsx`
+  - 27 frontend tests passed
+
+I also confirmed `git status --short` is clean after validation.
\ No newline at end of file
diff --git a/stages/007-simplify_gpt@1/status.json b/stages/007-simplify_gpt@1/status.json
new file mode 100644
index 000000000..67f1c0c4e
--- /dev/null
+++ b/stages/007-simplify_gpt@1/status.json
@@ -0,0 +1,6 @@
+{
+  "outcome": "succeeded",
+  "notes": "Stage completed: simplify_gpt",
+  "failure_reason": null,
+  "timestamp": "2026-05-24T02:25:22.173983Z"
+}
\ No newline at end of file
diff --git a/stages/008-verify@1/script_invocation.json b/stages/008-verify@1/script_invocation.json
new file mode 100644
index 000000000..7ad2687d7
--- /dev/null
+++ b/stages/008-verify@1/script_invocation.json
@@ -0,0 +1,5 @@
+{
+  "script": "git fetch origin main 2>&1 && git merge --no-edit --no-stat origin/main 2>&1 && cargo +nightly-2026-04-14 fmt --all 2>&1 && cargo dev docs refresh 2>&1 && cargo +nightly-2026-04-14 fmt --check --all 2>&1 && { command -v rg >/dev/null 2>&1 || { echo 'rg is required for verify'; exit 127; }; } && ! rg -n 'AuthMode::Disabled|RunAuthMethod|RunSubjectProvenance|\\bActorRef\\b|\\bActorKind\\b|AuthenticatedSubject|AuthenticatedService|AuthorizeRunScoped|AuthorizeRunBlob|AuthorizeStageArtifact|AuthorizeCommandLog|auth_method\\s*==\\s*\"disabled\"' lib/crates apps lib/packages docs/public/api-reference/fabro-api.yaml 2>&1 && cargo +nightly-2026-04-14 clippy --workspace --all-targets -- -D warnings 2>&1 && cargo nextest run --workspace --status-level slow --profile ci 2>&1 && cargo dev docs check 2>&1 && bun install --frozen-lockfile 2>&1 && (cd apps/fabro-web && bun run typecheck) 2>&1 && (cd apps/fabro-web && bun run test) 2>&1 && (cd lib/packages/fabro-api-client && bun run typecheck) 2>&1 && cargo dev build -- -p fabro-cli --release 2>&1",
+  "command": "exec 2>&1\ngit fetch origin main 2>&1 && git merge --no-edit --no-stat origin/main 2>&1 && cargo +nightly-2026-04-14 fmt --all 2>&1 && cargo dev docs refresh 2>&1 && cargo +nightly-2026-04-14 fmt --check --all 2>&1 && { command -v rg >/dev/null 2>&1 || { echo 'rg is required for verify'; exit 127; }; } && ! rg -n 'AuthMode::Disabled|RunAuthMethod|RunSubjectProvenance|\\bActorRef\\b|\\bActorKind\\b|AuthenticatedSubject|AuthenticatedService|AuthorizeRunScoped|AuthorizeRunBlob|AuthorizeStageArtifact|AuthorizeCommandLog|auth_method\\s*==\\s*\"disabled\"' lib/crates apps lib/packages docs/public/api-reference/fabro-api.yaml 2>&1 && cargo +nightly-2026-04-14 clippy --workspace --all-targets -- -D warnings 2>&1 && cargo nextest run --workspace --status-level slow --profile ci 2>&1 && cargo dev docs check 2>&1 && bun install --frozen-lockfile 2>&1 && (cd apps/fabro-web && bun run typecheck) 2>&1 && (cd apps/fabro-web && bun run test) 2>&1 && (cd lib/packages/fabro-api-client && bun run typecheck) 2>&1 && cargo dev build -- -p fabro-cli --release 2>&1",
+  "language": "shell"
+}
\ No newline at end of file