@bendyline/gilde 0.1.37 → 0.1.39

package/data/craftbook-templates/ci/citation-audit/versions/1.0.2/test.json

@@ -0,0 +1,152 @@
+{
+  "schemaVersion": 1,
+  "title": "Citation audit over seeded draft and source excerpts",
+  "objective": "Measure whether the citation-audit craftbook pairs claims to citations and reports supported, partial, unsupported, dead, and uncited claims.",
+  "tags": [
+    "corpus"
+  ],
+  "prompt": "In the `Citation Audit Eval` project, read only `source/draft.md` and `source/source-excerpts.md`, then write `tasks/eval/audit.md`. Audit exactly the four cited claims `Claim [A]` through `Claim [D]`, plus the uncited sentence about Finance receiving a weekly exception export. Do not add Claim [E], Claim [F], outside company metrics, or imagined source sections. Use verdicts Supported, Not-Supported, Source-Missing/Dead, and Uncited-Claim, with source quotes and concrete fixes.",
+  "setup": {
+    "projectName": "Citation Audit Eval",
+    "about": "Self-contained eval project for citation-audit. Draft and source excerpts are seeded under workspace/source.",
+    "missionObjectives": "Audit every citation and uncited factual claim with strict verdicts, quotes from source excerpts, and concrete fixes.",
+    "files": [
+      {
+        "path": "source/draft.md",
+        "content": "# Draft\n\nGuided returns intake reduced median first response from 18 hours to 6 hours [A]. The pilot covered every SKU in the catalog [B]. Customers were 34% more satisfied after photo evidence was required [C]. Finance now receives a weekly exception export. Warehouse teams need barcode exception training before national rollout [D].\n"
+      },
+      {
+        "path": "source/source-excerpts.md",
+        "content": "# Source excerpts\n\n[A] Pilot dashboard: median first response improved from 18 hours to 6 hours for the 18 pilot SKUs.\n[B] Rollout memo: March pilot covered 18 home-office accessory SKUs, not the full catalog.\n[C] Source missing: no satisfaction survey was run during the pilot.\n[D] Operations notes: warehouse teams need barcode exception training before national rollout.\n"
+      }
+    ],
+    "worker": {
+      "name": "Iris",
+      "role": "Reviewer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "workspace/audit.md audits all seeded citations and uncited claims with rubric verdicts, source quotes, counts, and fixes.",
+    "deliverables": [
+      {
+        "path": "tasks/eval/audit.md",
+        "kind": "markdown-report",
+        "minBytes": 1200,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Supported|Partially-Supported|Not-Supported|Source-Missing|Uncited-Claim",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "\\[A\\]|\\[B\\]|\\[C\\]|\\[D\\]",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Claim \\[A\\][\\s\\S]{0,500}Supported",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Claim \\[B\\][\\s\\S]{0,500}Not-Supported",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Claim \\[C\\][\\s\\S]{0,500}Source-Missing|Claim \\[C\\][\\s\\S]{0,500}Dead",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Claim \\[D\\][\\s\\S]{0,500}Supported",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Finance[\\s\\S]{0,500}Uncited-Claim|Uncited-Claim[\\s\\S]{0,500}Finance",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "18\\s*hours|6\\s*hours|18 home-office accessory SKUs|full catalog",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "34%|satisfaction survey",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Finance|weekly exception export",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "fix|correction|replace|add a citation",
+            "flags": "i"
+          },
+          {
+            "kind": "notContains",
+            "file": "tasks/eval/audit.md",
+            "pattern": "Claim \\[E\\]|Claim \\[F\\]|cost per return|50 concurrent|RESTful API|Q3|stress tests|Financial summary|System architecture|expansion of this report",
+            "flags": "i",
+            "label": "do not invent audit claims or source excerpts beyond the seeded draft"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/audit.md",
+      "kind": "markdown"
+    },
+    "axes": [
+      {
+        "name": "grounding",
+        "description": "Seeded facts appear faithfully and the source is cited."
+      },
+      {
+        "name": "structure",
+        "description": "Sections progress logically for the artifact class."
+      },
+      {
+        "name": "completeness",
+        "description": "Required elements (risks, next actions, scope) are all present."
+      },
+      {
+        "name": "tone",
+        "description": "The register fits the audience and artifact class."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "claim-citation pairing",
+    "strict verdicts",
+    "actionable fixes"
+  ]
+}

package/data/craftbook-templates/cl/cli-tool/versions/1.1.2/craftbook.json

@@ -0,0 +1,159 @@
+{
+  "id": "cli-tool",
+  "name": "Command-Line Tool",
+  "description": "Build a command-line tool with subcommands, flags, arguments, help text, and proper exit codes. Designs the CLI UX FIRST — the command surface, flags with defaults, positional args, help/usage output, and exit-code conventions — because a CLI's contract is its interface, then implements argument parsing and command logic, then writes tests for parsing, help, happy paths, and error exits. Covers argv parsing, subcommands, --flags and -short forms, --help, stdin/stdout/stderr conventions, and non-zero exit codes on error.\n\nA gallery craftbook generated from an archetype spec. It runs\n`phase → (per-phase gate) → … → evaluate → (loop) → finish`. Each build\nphase that produces a checkable artifact is followed by a **runtime\ngate-checkpoint** — the runtime verifies the artifact and routes with no\nmodel turn, looping back to redo the phase on a miss. The final `evaluate`\nstep holds a static deliverable gate plus a reviewer QA pass. What it adds\nover the generic `build-loop`: a specialist role per phase, a\ndomain-correct ordering, and a concrete per-phase quality bar.\n\nDeliverables marked \"artifact\" land in the project's artifacts drawer (`write_artifact` / `read_artifact`), not the shipped workspace — review output is not product source.\n\nPhases:\n\n1. CLI UX scope (planner) — lock the command surface, flags, args, exit codes → gated on artifact `{{workPath}}/ux-scope.md` (markdown-notes)\n2. Build the CLI (developer) — implement parsing + command logic → gated on `src/cli.js` (code-module)\n3. Write CLI tests (developer) — test parsing, help, happy paths, error exits → gated on `src/cli.test.js` (code-with-tests)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "ux-scope",
+  "triggers": [
+    "build a cli tool",
+    "command line tool",
+    "make a terminal command",
+    "cli with subcommands",
+    "argv parser"
+  ],
+  "steps": [
+    {
+      "id": "ux-scope",
+      "name": "CLI UX scope",
+      "description": "lock the command surface, flags, args, exit codes",
+      "prompt": "Design the command-line interface before any logic — the interface IS the contract. Step 1: Enumerate the commands/subcommands and what each does. Step 2: For each, list the positional arguments (required vs optional) and the flags (`--long`, `-s` short form, type, default value, whether it takes a value). Step 3: Draft the exact `--help`/usage text the tool prints, including a one-line synopsis per command. Step 4: Define the IO and exit-code contract: what goes to stdout vs stderr, exit 0 on success, distinct non-zero codes for usage errors vs runtime errors. Step 5: Write a numbered acceptance-criteria checklist of 5-9 items (e.g. 'running with no args or --help prints usage and exits 0', 'an unknown flag prints an error to stderr and exits non-zero', 'the primary command produces the documented output'). `write_task_note` the command spec + checklist and write the same to the produces path. No implementation yet.\n\nThe deliverable `{{workPath}}/ux-scope.md` lands in the project's artifacts drawer — write it with `write_artifact` and read it back with `read_artifact`; the shipped workspace stays untouched.",
+      "suggestedRole": "planner",
+      "advanceWhen": {
+        "file": "{{workPath}}/ux-scope.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/ux-scope.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/ux-scope.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "ux-scope",
+        "maxAttempts": 3
+      },
+      "next": "build"
+    },
+    {
+      "id": "build",
+      "name": "Build the CLI",
+      "description": "implement parsing + command logic",
+      "prompt": "Implement the CLI in a single runnable source file with a real entry point. Step 1: Parse argv into commands, positional args, and flags exactly as scoped — support both long and short flag forms and apply defaults. Step 2: Implement each command's logic; write normal output to stdout and diagnostics/errors to stderr. Step 3: Wire `--help` (and bare invocation) to print the usage text and exit 0; on unknown command/flag or missing required arg, print a clear error to stderr and exit non-zero. Step 4: Ensure the process exits with the documented code in every path. On a loop-back, fix only the named gaps without regressing working commands. `write_task_note` the file path and which criteria now pass.",
+      "suggestedRole": "developer",
+      "advanceWhen": {
+        "file": "src/cli.js",
+        "minBytes": 1,
+        "sniff": "nonempty"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "src/cli.js",
+            "bytes": 200
+          },
+          {
+            "kind": "sniff",
+            "file": "src/cli.js",
+            "sniff": "nonempty"
+          },
+          {
+            "kind": "esmImports",
+            "file": "src/cli.js"
+          },
+          {
+            "kind": "sourceParses",
+            "file": "src/cli.js"
+          }
+        ],
+        "onReject": "build",
+        "maxAttempts": 3
+      },
+      "next": "test"
+    },
+    {
+      "id": "test",
+      "name": "Write CLI tests",
+      "description": "test parsing, help, happy paths, error exits",
+      "prompt": "Write tests that invoke the CLI and inspect output + exit code. Step 1: Assert `--help` and bare invocation print usage and exit 0. Step 2: Assert each command's happy path produces the documented stdout and exits 0. Step 3: Assert error cases (unknown flag, missing required arg, bad value) print to stderr and exit non-zero with the right code. Step 4: Assert flag parsing handles long form, short form, and defaults. Capture stdout/stderr separately. `write_task_note` the test file path and pass/fail count.",
+      "suggestedRole": "developer",
+      "advanceWhen": {
+        "file": "src/cli.test.js",
+        "minBytes": 1,
+        "sniff": "nonempty"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "src/cli.js",
+            "bytes": 200
+          },
+          {
+            "kind": "sniff",
+            "file": "src/cli.js",
+            "sniff": "nonempty"
+          },
+          {
+            "kind": "esmImports",
+            "file": "src/cli.js"
+          },
+          {
+            "kind": "sourceParses",
+            "file": "src/cli.js"
+          }
+        ],
+        "onReject": "test",
+        "maxAttempts": 4
+      },
+      "next": "evaluate"
+    },
+    {
+      "id": "evaluate",
+      "name": "Evaluate",
+      "description": "Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.",
+      "prompt": "Run the tests and invoke the CLI directly. For EACH acceptance criterion: confirm --help/usage works and exits 0, confirm each command's output matches the spec, confirm errors go to stderr with a non-zero exit, and confirm flag parsing (long, short, defaults) works. Write PASS/FAIL per criterion.\n\nThen route — this is the whole point of the loop:\n\n- **Every criterion PASSES →** call `advance_task_step({ ref, stepId: \"evaluate\", next: \"finish\" })`.\n- **Any criterion FAILS →** write the specific gaps to notes, then call `advance_task_step({ ref, stepId: \"evaluate\", next: \"build\" })` to loop back. The builder fixes exactly those gaps.\n\nNever route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in.",
+      "suggestedRole": "reviewer",
+      "consumes": [
+        {
+          "file": "src/cli.js"
+        }
+      ],
+      "next": "build"
+    },
+    {
+      "id": "finish",
+      "name": "Finish",
+      "description": "All acceptance criteria met. Stamp a short summary and report DONE.",
+      "prompt": "Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.",
+      "suggestedRole": "developer",
+      "terminal": true
+    }
+  ],
+  "version": "1.1.2",
+  "releasedAt": "2026-08-21T00:00:00Z",
+  "paramSchema": {
+    "type": "object",
+    "properties": {
+      "workPath": {
+        "type": "string",
+        "title": "Working folder",
+        "description": "Per-task working folder in the artifacts drawer. Defaults to this task's own folder so runs never collide; override with a stable name when you deliberately want runs to share files.",
+        "default": "{{task.dir}}"
+      }
+    }
+  },
+  "minGezelVersion": "1.26233"
+}

package/data/craftbook-templates/cl/cli-tool/versions/1.1.2/test.json

@@ -0,0 +1,131 @@
+{
+  "schemaVersion": 1,
+  "title": "Small CSV summary CLI",
+  "objective": "Measure whether the cli-tool craftbook scopes command UX, writes a runnable Node CLI, and includes tests/examples.",
+  "tags": [
+    "html-page"
+  ],
+  "prompt": "In the `CLI Tool Eval` project, build a small dependency-free Node ESM CLI named `summarize-orders`. Write `src/cli.js` and `src/cli.test.js`. The CLI should accept `--input <csv>`, `--format table|json`, and `--min-total <number>`, parse order rows with customer,status,total fields, filter by minimum total, and print either a text table or JSON summary. The test file should use Node assert or a tiny test harness and cover JSON output, table output, and min-total filtering.",
+  "setup": {
+    "projectName": "CLI Tool Eval",
+    "about": "Self-contained eval project for cli-tool. Deliverables are workspace/src/cli.js and workspace/src/cli.test.js.",
+    "missionObjectives": "Produce a small dependency-free Node CLI with argument parsing, CSV parsing, JSON/table output, and tests.",
+    "files": [],
+    "worker": {
+      "name": "Koen",
+      "role": "Developer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "src/cli.js implements argument parsing/output modes and src/cli.test.js covers the core CLI behavior.",
+    "deliverables": [
+      {
+        "path": "src/cli.js",
+        "kind": "code-module",
+        "minBytes": 1600,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "src/cli.js",
+            "pattern": "process\\.argv|argv",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.js",
+            "pattern": "--input",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.js",
+            "pattern": "--format",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.js",
+            "pattern": "--min-total",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.js",
+            "pattern": "read_file|fs",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.js",
+            "pattern": "JSON\\.stringify|console\\.log",
+            "flags": "i"
+          }
+        ]
+      },
+      {
+        "path": "src/cli.test.js",
+        "kind": "code-with-tests",
+        "minBytes": 900,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "src/cli.test.js",
+            "pattern": "assert|describe|it|test",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.test.js",
+            "pattern": "json",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.test.js",
+            "pattern": "table",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "src/cli.test.js",
+            "pattern": "min-total|filter",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "src/cli.js",
+      "kind": "typescript"
+    },
+    "axes": [
+      {
+        "name": "structure",
+        "description": "Layout and hierarchy fit the content; sections are scannable and coherent."
+      },
+      {
+        "name": "grounding",
+        "description": "Seeded source content is used faithfully — no invented facts or dropped requirements."
+      },
+      {
+        "name": "interactivity",
+        "description": "Controls and states the page claims to offer actually work."
+      },
+      {
+        "name": "polish",
+        "description": "Styling is intentional and responsive rather than unstyled or broken."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "CLI UX scope",
+    "argument parsing",
+    "test coverage"
+  ]
+}

package/data/craftbook-templates/co/codebase-refactoring-review/versions/1.0.2/craftbook.json

@@ -0,0 +1,165 @@
+{
+  "id": "codebase-refactoring-review",
+  "name": "Codebase Refactoring Review",
+  "description": "Review an entire codebase for the least cogent, highest-leverage refactoring opportunities without changing source. Uses the repository map and per-file AI health index to rank weak areas, then verifies each lead in the real code and leaves a file-and-line-cited morning report with sequenced refactor plans, risk, effort, and tests that must protect behavior. Use for a weekly code-health sweep, refactoring review, technical-debt triage, finding code smells, or deciding which files should be refactored first.\n\nA gallery craftbook generated from an archetype spec. It runs\n`phase → (per-phase gate) → … → evaluate → (loop) → finish`. Each build\nphase that produces a checkable artifact is followed by a **runtime\ngate-checkpoint** — the runtime verifies the artifact and routes with no\nmodel turn, looping back to redo the phase on a miss. The final `evaluate`\nstep holds a static deliverable gate plus a reviewer QA pass. What it adds\nover the generic `build-loop`: a specialist role per phase, a\ndomain-correct ordering, and a concrete per-phase quality bar.\n\nDeliverables marked \"artifact\" land in the project's artifacts drawer (`write_artifact` / `read_artifact`), not the shipped workspace — review output is not product source.\n\nPhases:\n\n1. Rank weak areas (reviewer) — use the repo map and per-file health index to build a coverage-aware candidate list → gated on artifact `{{workPath}}/refactoring/candidates.md` (markdown-notes)\n2. Verify refactoring candidates (reviewer) — inspect candidate symbols, responsibilities, callers, and tests in real source → gated on artifact `{{workPath}}/refactoring/verified.md` (markdown-notes)\n3. Write the refactoring report (reviewer) — prioritized code-health report with safe, sequenced refactor plans → gated on artifact `{{workPath}}/refactoring-review.md` (markdown-report)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "rank",
+  "triggers": [
+    "codebase refactoring review",
+    "what should we refactor",
+    "find the least cogent code",
+    "weekly code health review",
+    "technical debt sweep",
+    "nightly refactoring review"
+  ],
+  "toolsets": [
+    {
+      "toolsetId": "builtin.code-intel",
+      "autoAllow": true,
+      "reason": "rank indexed file-health leads and inspect only the relevant symbols"
+    },
+    {
+      "toolsetId": "builtin.git",
+      "autoAllow": true,
+      "reason": "read-only history and churn evidence for prioritizing refactors"
+    }
+  ],
+  "steps": [
+    {
+      "id": "rank",
+      "name": "Rank weak areas",
+      "description": "use the repo map and per-file health index to build a coverage-aware candidate list",
+      "prompt": "Start with `map_repo` and record the indexed review coverage (reviewed/eligible), average health, major-issue count, and worst files. Call `list_file_issues` across the repo, then use `search_code` for cross-cutting duplication, oversized orchestration, unclear ownership, error handling, and tightly coupled state. If git history is available, use read-only `run_git` queries to add churn/fix-frequency evidence; missing history is not a failure. Rank candidate files or cohesive areas using explicit factors: low indexed health, confirmed major issues, too many responsibilities, coupling/fan-out, duplication, change frequency, test difficulty, and blast radius. Do not call an indexed lead a finding yet and do not modify source. Write the coverage statement, ranking method, and top candidates to `{{workPath}}/refactoring/candidates.md`.\n\nThe deliverable `{{workPath}}/refactoring/candidates.md` lands in the project's artifacts drawer — write it with `write_artifact` and read it back with `read_artifact`; the shipped workspace stays untouched.",
+      "suggestedRole": "reviewer",
+      "advanceWhen": {
+        "file": "{{workPath}}/refactoring/candidates.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/refactoring/candidates.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/refactoring/candidates.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "rank",
+        "maxAttempts": 3
+      },
+      "next": "verify"
+    },
+    {
+      "id": "verify",
+      "name": "Verify refactoring candidates",
+      "description": "inspect candidate symbols, responsibilities, callers, and tests in real source",
+      "prompt": "Verify every shortlisted candidate in source. Call `file_review` for the indexed rationale, `outline_file` before reading a large file, `read_symbol` for the implicated units, and `find_references` to understand callers and blast radius. Locate tests with `search_code`/`grep_files`. For each confirmed candidate record real file:line evidence, the current responsibilities, the specific cohesion/coupling/duplication/testability problem, why it matters, a bounded refactor seam, dependencies and migration order, behavior-preserving tests required first, effort, risk, and expected payoff. Reject false-positive index leads explicitly. Prefer a smaller ranked list of verified opportunities over a long smell inventory. Write the evidence matrix to `{{workPath}}/refactoring/verified.md`; do not change code.\n\nThe deliverable `{{workPath}}/refactoring/verified.md` lands in the project's artifacts drawer — write it with `write_artifact` and read it back with `read_artifact`; the shipped workspace stays untouched.",
+      "suggestedRole": "reviewer",
+      "advanceWhen": {
+        "file": "{{workPath}}/refactoring/verified.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/refactoring/verified.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/refactoring/verified.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "verify",
+        "maxAttempts": 3
+      },
+      "next": "report"
+    },
+    {
+      "id": "report",
+      "name": "Write the refactoring report",
+      "description": "prioritized code-health report with safe, sequenced refactor plans",
+      "prompt": "Write `{{workPath}}/refactoring-review.md` for an admin to read in the morning. Use these exact sections: `## Executive summary`, `## Index coverage`, `## Ranked opportunities`, `## Cross-cutting themes`, `## Defer or leave alone`, and `## Recommended sequence`. Each ranked opportunity must cite real file:line evidence; state why the area is insufficiently cogent; name the proposed seam and steps; list behavior-preserving tests; and include impact, effort, risk, and dependencies. Separate systemic themes from individual files. State index gaps and unreviewed areas so a quiet report is not mistaken for a clean codebase. Do not include style-only churn or recommend rewrites without a migration boundary. On a loop-back, repair only the named report gaps.\n\nThe deliverable `{{workPath}}/refactoring-review.md` lands in the project's artifacts drawer — write it with `write_artifact` and read it back with `read_artifact`; the shipped workspace stays untouched.",
+      "suggestedRole": "reviewer",
+      "advanceWhen": {
+        "file": "{{workPath}}/refactoring-review.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/refactoring-review.md",
+            "bytes": 1800,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/refactoring-review.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i",
+            "label": "at least one markdown heading",
+            "artifact": true
+          }
+        ],
+        "onReject": "report",
+        "maxAttempts": 4
+      },
+      "next": "evaluate"
+    },
+    {
+      "id": "evaluate",
+      "name": "Evaluate",
+      "description": "Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.",
+      "prompt": "Grade `{{workPath}}/refactoring-review.md`: (1) index coverage and limitations are explicit; (2) every ranked item was verified in source and cites a real file:line; (3) ranking uses impact, evidence, effort, risk, and blast radius rather than file size alone; (4) each recommendation names a bounded seam, sequence, and behavior-preserving tests; (5) systemic themes are distinguished from local smells; (6) false positives and areas to leave alone are recorded; (7) no source was changed and no vague rewrite advice remains. PASS every item or loop back with exact gaps.\n\nThe deliverable lives in the project's artifacts drawer — open `{{workPath}}/refactoring-review.md` with `read_artifact`, not `read_file`.\n\nThen route — this is the whole point of the loop:\n\n- **Every criterion PASSES →** call `advance_task_step({ ref, stepId: \"evaluate\", next: \"finish\" })`.\n- **Any criterion FAILS →** write the specific gaps to notes, then call `advance_task_step({ ref, stepId: \"evaluate\", next: \"report\" })` to loop back. The builder fixes exactly those gaps.\n\nNever route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in.",
+      "suggestedRole": "reviewer",
+      "consumes": [
+        {
+          "file": "{{workPath}}/refactoring-review.md",
+          "artifact": true
+        }
+      ],
+      "next": "report"
+    },
+    {
+      "id": "finish",
+      "name": "Finish",
+      "description": "All acceptance criteria met. Stamp a short summary and report DONE.",
+      "prompt": "Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.",
+      "suggestedRole": "developer",
+      "terminal": true
+    }
+  ],
+  "version": "1.0.2",
+  "releasedAt": "2026-08-21T00:00:00Z",
+  "paramSchema": {
+    "type": "object",
+    "properties": {
+      "workPath": {
+        "type": "string",
+        "title": "Working folder",
+        "description": "Per-task working folder in the artifacts drawer. Defaults to this task's own folder so runs never collide; override with a stable name when you deliberately want runs to share files.",
+        "default": "{{task.dir}}"
+      }
+    }
+  },
+  "minGezelVersion": "1.26233"
+}