@bendyline/gilde 0.1.38 → 0.1.39

package/data/craftbook-templates/mo/monthly-recording-review/versions/1.0.2/test.json

@@ -0,0 +1,120 @@
+{
+  "schemaVersion": 1,
+  "title": "Monthly Recording Review — July",
+  "objective": "Measure whether the monthly-recording-review craftbook assesses every recording in the index against last month's marks and lands on exactly one next-month focus.",
+  "tags": [
+    "corpus"
+  ],
+  "prompt": "end of the month — can you go thru the recordings index and my notes and write the review? tasks/eval/report.md please",
+  "setup": {
+    "projectName": "Piano Practice",
+    "about": "Music practice project. Monthly practice recordings are indexed in workspace/recordings/index.json with a listening-notes file per recording; past monthly reviews live under workspace/reviews.",
+    "missionObjectives": "Once a month, review the recordings against last month's marks — proof over impression — and pick one focus for next month. Grounded in the seeded index and notes; no live services, no real credentials.",
+    "files": [
+      {
+        "path": "recordings/index.json",
+        "content": "[\n  {\n    \"file\": \"2026-07-05-nocturne.mp3\",\n    \"date\": \"2026-07-05\",\n    \"piece\": \"Chopin Nocturne Op. 9 No. 2\",\n    \"tasks/eval\": \"recordings/2026-07-05-nocturne.notes.md\"\n  },\n  {\n    \"file\": \"2026-07-12-nocturne.mp3\",\n    \"date\": \"2026-07-12\",\n    \"piece\": \"Chopin Nocturne Op. 9 No. 2\",\n    \"tasks/eval\": \"recordings/2026-07-12-nocturne.notes.md\"\n  },\n  {\n    \"file\": \"2026-07-19-bach-invention.mp3\",\n    \"date\": \"2026-07-19\",\n    \"piece\": \"Bach Invention No. 8\",\n    \"tasks/eval\": \"recordings/2026-07-19-bach-invention.notes.md\"\n  },\n  {\n    \"file\": \"2026-07-26-nocturne-full.mp3\",\n    \"date\": \"2026-07-26\",\n    \"piece\": \"Chopin Nocturne Op. 9 No. 2 (full run-through)\",\n    \"tasks/eval\": \"recordings/2026-07-26-nocturne-full.notes.md\"\n  }\n]\n"
+      },
+      {
+        "path": "recordings/2026-07-05-nocturne.notes.md",
+        "content": "# Listening notes — 2026-07-05 nocturne\n\nTempo 88 bpm. Left hand steadier than June — the arpeggios no longer bulge on beat three. Bar 12 ornaments still uneven, rushed on the second turn. Pedal is cleaner in the opening eight bars. Bar 32 run attempted at tempo: fell apart, restarted.\n"
+      },
+      {
+        "path": "recordings/2026-07-12-nocturne.notes.md",
+        "content": "# Listening notes — 2026-07-12 nocturne\n\nTempo 92 bpm. Bar 32 run holds at 85 in isolation but breaks in context. Dynamics are starting to appear — the first real crescendo shaping in bars 17-24. Ending chord voicing much better than the June recordings.\n"
+      },
+      {
+        "path": "recordings/2026-07-19-bach-invention.notes.md",
+        "content": "# Listening notes — 2026-07-19 bach invention\n\nFirst hands-together recording. 66 bpm, down from the 72 achieved hands-separate — expected. Evenness between the hands is good for a first attempt. Ornaments omitted for now. No memory slips.\n"
+      },
+      {
+        "path": "recordings/2026-07-26-nocturne-full.notes.md",
+        "content": "# Listening notes — 2026-07-26 nocturne full run-through\n\nFirst complete run-through ever recorded. 90 bpm working tempo. One stumble at bar 32 (recovered without stopping — new). Rubato in the middle section sounds intentional for the first time. The June recordings had no complete take at all.\n"
+      },
+      {
+        "path": "reviews/2026-06.md",
+        "content": "# Monthly review — June 2026\n\n## Marks\n\n- Chopin Nocturne Op. 9 No. 2 — 84 bpm working tempo, no complete take; left hand bulges on beat three; bar 32 run avoided entirely.\n- Bach Invention No. 8 — hands separate only, 72 bpm.\n\n## Focus for July\n\nOne focus: left-hand evenness in the nocturne. Everything else is secondary.\n"
+      }
+    ],
+    "worker": {
+      "name": "Iris",
+      "role": "Coach"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "tasks/eval/report.md assesses every indexed recording with concrete evidence, states deltas against the June review, gives a verdict on June's focus, and names exactly one focus for August.",
+    "deliverables": [
+      {
+        "path": "tasks/eval/report.md",
+        "kind": "markdown-report",
+        "minBytes": 1000,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i"
+          },
+          {
+            "kind": "valuesSubsetOf",
+            "file": "tasks/eval/report.md",
+            "sourceFiles": [
+              "recordings/index.json"
+            ],
+            "pattern": "(\\d{4}-\\d{2}-\\d{2}-[a-z-]+\\.mp3)",
+            "flags": "i",
+            "minMatches": 2,
+            "label": "recordings referenced by their real filenames"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "focus",
+            "flags": "i",
+            "label": "next-month focus named"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "better|steadier|cleaner|slower|improved|regressed",
+            "flags": "i",
+            "label": "delta language versus last month"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/report.md",
+      "kind": "markdown"
+    },
+    "axes": [
+      {
+        "name": "honesty",
+        "description": "Progress and regressions are both stated plainly — no flattery, no doom."
+      },
+      {
+        "name": "specificity",
+        "description": "Observations name passages, tempos, and habits a teacher would point at — never a bare 'sounds better'."
+      },
+      {
+        "name": "comparison rigor",
+        "description": "Every piece gets a delta against the June marks, and June's focus (left-hand evenness) gets an explicit verdict."
+      },
+      {
+        "name": "encouragement",
+        "description": "The review makes the player want to practice tomorrow, and lands on exactly one focus for next month."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "evidence over impression",
+    "month-over-month deltas",
+    "single focus"
+  ]
+}

package/data/craftbook-templates/mo/moodboard/versions/1.0.2/craftbook.json

@@ -0,0 +1,121 @@
+{
+  "id": "moodboard",
+  "name": "Visual Moodboard",
+  "description": "Compose a visual moodboard that captures a direction — palette, textures, type vibe, imagery, and reference tiles — from a written brief, so a team can align on a look before any production work begins. Lock the brief and the explicit dimensions of the mood (e.g. 'warm/editorial/handmade' vs 'cool/technical/precise') FIRST, then assemble a coherent grid of generated or sourced tiles plus a swatch strip and keyword callouts. Covers moodboards, style boards, visual direction, design inspiration, brand mood, and look-and-feel exploration.\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. Moodboard brief (visual-designer) — lock the mood dimensions, palette, and tile plan → gated on artifact `{{workPath}}/brief.md` (markdown-notes)\n2. Assemble the board (image-generator) — compose the coherent tile grid into one deliverable → gated on `index.html` (html-page)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "brief",
+  "triggers": [
+    "make a moodboard",
+    "create a style board",
+    "visual direction board",
+    "design inspiration board",
+    "look and feel board"
+  ],
+  "steps": [
+    {
+      "id": "brief",
+      "name": "Moodboard brief",
+      "description": "lock the mood dimensions, palette, and tile plan",
+      "prompt": "1) Distill the direction into 3-5 mood keywords (e.g. warm, editorial, handmade, calm). 2) Lock a palette with hex swatches and the type vibe (serif/sans, weight, feeling). 3) Plan the tile grid: enumerate the tiles you'll include — e.g. 2-3 imagery tiles, a texture tile, a typography specimen tile, a palette strip — and what each should evoke. 4) Note explicit anti-references (what this is NOT) so the board has a point of view. 5) Write an acceptance-criteria checklist: 'board expresses the 3-5 keywords', 'palette swatches show exact hex', 'a typography specimen is present', 'tiles are cohesive (one direction, not a scrapbook)', 'at least 6 tiles'. 6) Record brief + checklist via write_task_note AND to {{workPath}}/brief.md. No assembly yet.\n\nThe deliverable `{{workPath}}/brief.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": "visual-designer",
+      "advanceWhen": {
+        "file": "{{workPath}}/brief.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/brief.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/brief.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "brief",
+        "maxAttempts": 3
+      },
+      "next": "assemble"
+    },
+    {
+      "id": "assemble",
+      "name": "Assemble the board",
+      "description": "compose the coherent tile grid into one deliverable",
+      "prompt": "1) Read brief.md. 2) Generate or compose each planned tile to express the locked keywords and palette — imagery tiles via the image-generation capability (or curated solid/gradient/texture blocks if no image model is available), a swatch strip rendering the exact hex values, and a type specimen tile showing the chosen pairing with sample words. 3) Assemble everything into ONE deliverable: a single index.html grid (preferred — easy to render and tweak) laying out at least 6 cohesive tiles with the keyword list as a caption. 4) Keep the direction tight: every tile should feel like the same world. 5) Reference any generated files from assets/ if produced. 6) write_task_note the board path and tile count. Good looks like: someone could brief a designer from this board alone.",
+      "suggestedRole": "image-generator",
+      "advanceWhen": {
+        "file": "index.html",
+        "minBytes": 1,
+        "sniff": "html-complete"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "index.html",
+            "bytes": 800
+          },
+          {
+            "kind": "htmlLint",
+            "file": "index.html"
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkHtmlComplete",
+            "scope": "standard",
+            "inputs": {
+              "file": "index.html"
+            }
+          }
+        ],
+        "onReject": "assemble",
+        "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": "1) Open index.html and confirm it renders a grid of at least 6 tiles (PASS/FAIL). 2) Check it expresses the 3-5 mood keywords from brief.md — is the direction legible and cohesive, not a random scrapbook (PASS/FAIL)? 3) Confirm a palette swatch strip shows the exact hex values and a typography specimen is present. 4) Verify any referenced image files actually load. 5) Write PASS/FAIL per criterion. Incoherent/off-brief → loop back to assemble.\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: \"assemble\" })` 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": "index.html"
+        }
+      ],
+      "next": "assemble"
+    },
+    {
+      "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"
+}

package/data/craftbook-templates/mo/moodboard/versions/1.0.2/test.json

@@ -0,0 +1,106 @@
+{
+  "schemaVersion": 1,
+  "title": "Visual Moodboard smoke eval",
+  "objective": "Self-contained smoke eval for the Visual Moodboard craftbook using the media generic harness.",
+  "tags": [
+    "media"
+  ],
+  "prompt": "Our media files are all listd in source/media-index.csv. Can you put together the production plan for them as media-plan.json?",
+  "setup": {
+    "projectName": "Visual Moodboard Eval",
+    "about": "Self-contained eval project for moodboard. Seeded inputs are under workspace/source or workspace/fixtures; final deliverable is workspace/media-plan.json.",
+    "missionObjectives": "Use the Visual Moodboard craftbook/template, read the seeded local fixtures, and write media-plan.json without network calls, real credentials, or live services.",
+    "files": [
+      {
+        "path": "source/brief.md",
+        "content": "# Visual Moodboard Eval Brief\n\nClient: Boreal Desk, a home-office accessories company.\nAudience: operations leads who need an artifact they can use this week.\n\nFixed source facts for grounding:\n- The returns desk pilot covered 18 SKUs.\n- Median first response improved from 18 hours to 6 hours.\n- Preventable refund leakage fell from 14.2% to 8.9%.\n- The top unresolved complaint is status silence after photo submission.\n- Required next actions are automated status emails, barcode-exception training, and a weekly Finance exception export.\n\nUse these facts when the task asks for prose, analysis, copy, UI content, or test data. Do not use live web services, real credentials, or current outside data.\n\nCraftbook under test: moodboard - Visual Moodboard.\n"
+      },
+      {
+        "path": "source/media-index.csv",
+        "content": "filename,type,duration_or_size,notes\npilot-interview.wav,audio,00:02:15,mentions response time and status silence\nreturns-desk.png,image,1600x900,shows intake form and photo evidence step\nlaunch-clip.mp4,video,00:01:05,shows barcode exception handling\n"
+      }
+    ],
+    "worker": {
+      "name": "Jules",
+      "role": "Producer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "media-plan.json is a valid JSON media-production manifest grounded in seeded fixture metadata.",
+    "deliverables": [
+      {
+        "path": "media-plan.json",
+        "kind": "json",
+        "minBytes": 700,
+        "checks": [
+          {
+            "kind": "sniff",
+            "file": "media-plan.json",
+            "sniff": "json-valid"
+          },
+          {
+            "kind": "contains",
+            "file": "media-plan.json",
+            "pattern": "pilot-interview\\.wav|returns-desk\\.png|launch-clip\\.mp4",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "media-plan.json",
+            "pattern": "outputs|assets|steps|cues|verification",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "media-plan.json",
+            "pattern": "accessibility|alt|caption|transcript|metadata",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "media-plan.json",
+      "kind": "json"
+    },
+    "axes": [
+      {
+        "name": "completeness",
+        "description": "The manifest covers sources, outputs, and processing steps end to end."
+      },
+      {
+        "name": "grounding",
+        "description": "Only seeded media fixtures are referenced, with faithful metadata."
+      },
+      {
+        "name": "accessibility",
+        "description": "Alt text, captions, or transcripts are planned wherever the class requires."
+      },
+      {
+        "name": "precision",
+        "description": "Timestamps, crops, and cues are specific enough to execute."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "media fixture handling",
+    "structured manifest",
+    "verification metadata"
+  ],
+  "extensions": {
+    "legacySimulators": [
+      {
+        "id": "moodboard-media-fixtures",
+        "kind": "data-source",
+        "status": "implemented",
+        "description": "Seeded local media index replacing real audio, video, image generation, and external media metadata services."
+      }
+    ]
+  }
+}

package/data/craftbook-templates/mo/morning-brief/versions/1.0.2/craftbook.json

@@ -0,0 +1,119 @@
+{
+  "id": "morning-brief",
+  "name": "Morning Topic Brief",
+  "description": "Be informed without the doomscroll: this morning's items per watched topic, each with its source named, rumor separated from reporting, and yesterday's noise skipped — five minutes that replace an hour of feeds. Locks the brief rules FIRST, then gathers from the feeds, then writes the dated brief into the archive next to the previous ones. Use for morning briefs, daily topic digests, 'what happened overnight', news catch-ups on watched topics, and staying informed without the scroll.\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. Set the brief rules (editor) — topics read, yesterday skimmed, brief rules locked → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Gather the morning's items (researcher) — fetch per topic, annotate sources, mark the cuts → gated on artifact `{{workPath}}/items.md` (markdown-notes)\n3. Write today's brief (copywriter) — the dated brief in the archive, sources named, rumor flagged\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "morning brief",
+    "what happened overnight",
+    "today's brief",
+    "run the morning digest",
+    "catch me up on my topics"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Set the brief rules",
+      "description": "topics read, yesterday skimmed, brief rules locked",
+      "prompt": "Set the brief's rules before touching the feeds. Step 1: read topics.json from the workspace — the topics being watched and their names — and skim the most recent brief in briefs/ so you know what yesterday already covered. Step 2: lock the brief rules — one section per topic; every item names its source, with a link when the feed gives one; an item that already appeared in yesterday's brief is skipped or explicitly marked as already covered, never re-reported as new; an item resting on a single source is flagged as rumor or unconfirmed, never stated as fact. Step 3: write an acceptance-criteria checklist ('a section for every topic in topics.json', 'every item cites its source', 'no yesterday item re-reported as new', 'single-source items flagged', 'five minutes to read, not an hour'). Call write_task_note with the topic list + checklist and write {{workPath}}/scope.md. No fetching yet.\n\nThe deliverable `{{workPath}}/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": "editor",
+      "advanceWhen": {
+        "file": "{{workPath}}/scope.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/scope.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/scope.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "scope",
+        "maxAttempts": 3
+      },
+      "next": "gather"
+    },
+    {
+      "id": "gather",
+      "name": "Gather the morning's items",
+      "description": "fetch per topic, annotate sources, mark the cuts",
+      "prompt": "Gather this morning's items. Step 1: pull the current items for each watched topic — through the project's feed capabilities when the feeds are wired live, or from the freshest feed snapshots in the workspace. Step 2: for each topic, list every fetched item with its title, source, link, and anything the feed says about its confidence. Step 3: mark the cuts — which items are duplicates of yesterday's brief, which rest on a single source, which are noise not worth the owner's morning — and say why in a few words each. Step 4: write the full annotated item list to {{workPath}}/items.md and call write_task_note with the fetched/kept/cut counts per topic.\n\nThe deliverable `{{workPath}}/items.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": "researcher",
+      "advanceWhen": {
+        "file": "{{workPath}}/items.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/items.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/items.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "gather",
+        "maxAttempts": 3
+      },
+      "next": "brief"
+    },
+    {
+      "id": "brief",
+      "name": "Write today's brief",
+      "description": "the dated brief in the archive, sources named, rumor flagged",
+      "prompt": "Write today's brief from the gathered items. Step 1: date the brief and write it into the archive — a new file in briefs/ named for today (YYYY-MM-DD.md), next to the previous ones. Step 2: one section per topic, using the topic's name as the heading; under it, the kept items — each a sentence or two of what actually happened, with the source named and linked where the feed gave one. Step 3: keep rumor separate — a single-sourced item appears with its flag ('unconfirmed', 'single source'), never dressed as settled reporting. Step 4: where yesterday's story is still the story, one line saying nothing new beats re-reporting it. Step 5: five minutes of reading, no filler. Call write_task_note with the item count per topic and the brief's path.",
+      "suggestedRole": "copywriter",
+      "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": "Open today's brief in briefs/ and {{workPath}}/items.md and verify every criterion from {{workPath}}/scope.md: a section for every topic in topics.json, every item citing a named source (linked where the feed gave one), no item from yesterday's brief re-reported as new, every single-sourced item carrying its rumor or unconfirmed flag, and total reading time near five minutes. Cross-check the brief's items against the gathered list — nothing in the brief that was not fetched. Write PASS/FAIL per criterion; on any failure, name the gap and loop back to brief.\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: \"brief\" })` to loop back. The builder fixes exactly those gaps.\n\nNever route to `finish` while any criterion is unmet. If you advance without a target, the loop sends you back by design. After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in.",
+      "suggestedRole": "reviewer",
+      "next": "brief"
+    },
+    {
+      "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"
+}