@bendyline/gilde 0.1.37 → 0.1.39

package/data/craftbook-templates/an/annotated-bibliography/versions/1.0.2/craftbook.json

@@ -0,0 +1,158 @@
+{
+  "id": "annotated-bibliography",
+  "name": "Annotated Bibliography",
+  "description": "Build a formal annotated bibliography for a research topic: a planner locks the scope, the citation style, and the annotation rubric (summary + evaluation + relevance), then a researcher locates each source and writes a consistent, properly-formatted annotation that summarizes, critically evaluates, and situates it for the research question. The ordering matters because annotations written without a fixed rubric vary wildly in depth and a bibliography assembled without a citation style is inconsistent; locking style and the three-part annotation rubric first yields a uniform, scholarly, reusable artifact. Use this for annotated bibliographies, a sourced reference list with summaries, research source logs, literature foundations for a paper, and 'list and evaluate my sources' requests.\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. Scope the bibliography (planner) — topic, citation style, and annotation rubric → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Locate & evaluate sources (researcher) — find each source, draft its three-part annotation → gated on artifact `{{workPath}}/sources.md` (markdown-notes)\n3. Write the bibliography (copywriter) — formatted, ordered, uniformly-annotated bibliography → gated on `bibliography.md` (markdown-doc)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "annotated bibliography",
+    "list and evaluate sources",
+    "sourced reference list with summaries",
+    "research source log",
+    "bibliography with annotations"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope the bibliography",
+      "description": "topic, citation style, and annotation rubric",
+      "prompt": "1. State the research question/topic and the inclusion criteria (source types, recency, minimum count — aim 6+). 2. Lock the citation style exactly (APA, MLA, Chicago, or a stated custom format) with a one-line example of a correctly-formatted entry. 3. Lock the annotation rubric: each annotation must have (a) a 2-4 sentence summary of the source, (b) a critical evaluation of its credibility/method/bias, and (c) a relevance note tying it to the research question. 4. Decide ordering (alphabetical by author, or thematic). 5. Write an acceptance-criteria checklist (e.g. 'every entry is formatted in the locked citation style', 'every annotation has all three rubric parts', 'sources meet the inclusion criteria', 'entries are ordered as specified', 'minimum source count met'). 6. Record decisions. Call write_task_note with the style example, rubric, and checklist; write the same to {{workPath}}/scope.md.\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": "planner",
+      "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": "research"
+    },
+    {
+      "id": "research",
+      "name": "Locate & evaluate sources",
+      "description": "find each source, draft its three-part annotation",
+      "prompt": "1. Find sources meeting the inclusion criteria, prioritizing primary and authoritative works on the topic. 2. For each, record in {{workPath}}/sources.md: the full bibliographic data needed for the locked citation style, plus a draft three-part annotation (summary, critical evaluation, relevance) per the rubric. 3. In the evaluation part, judge the source honestly — author authority, method or evidence quality, recency, and any bias or limitation — not just praise. 4. Verify each source actually exists and the bibliographic details are correct; flag any you cannot confirm. 5. Ensure you exceed the minimum count and cover the topic without padding with weak sources. Good looks like a sources file where each entry already has correct citation data and a complete three-part annotation. Do not format the final document yet.\n\nThe deliverable `{{workPath}}/sources.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}}/sources.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/sources.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/sources.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "research",
+        "maxAttempts": 3
+      },
+      "next": "write"
+    },
+    {
+      "id": "write",
+      "name": "Write the bibliography",
+      "description": "formatted, ordered, uniformly-annotated bibliography",
+      "prompt": "1. Write bibliography.md as an ordered list of entries: each a citation formatted exactly in the locked style, followed by its annotation containing all three rubric parts (summary, evaluation, relevance). 2. Apply the citation style precisely and consistently — punctuation, author order, italics/markers as the style requires. 3. Keep annotation length and depth uniform across entries. 4. Order entries as specified (alphabetical/thematic). 5. Draw everything from {{workPath}}/sources.md; do not invent or alter citation data. 6. On a loop-back, fix only the named gaps. 7. Call write_task_note with the bibliography path and entry count.\n\nBefore working, open the artifact input `{{workPath}}/sources.md` with `read_artifact`; do not look for it in the workspace.",
+      "suggestedRole": "copywriter",
+      "consumes": [
+        {
+          "file": "{{workPath}}/sources.md",
+          "artifact": true
+        }
+      ],
+      "advanceWhen": {
+        "file": "bibliography.md",
+        "minBytes": 1,
+        "sniff": "nonempty"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "bibliography.md",
+            "bytes": 500
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkContains",
+            "scope": "standard",
+            "inputs": {
+              "file": "bibliography.md",
+              "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+              "flags": "i"
+            }
+          }
+        ],
+        "onReject": "write",
+        "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": "Open bibliography.md and check EACH criterion PASS/FAIL: (1) every entry is formatted in the locked citation style and is consistent (check 3 against the style example in {{workPath}}/scope.md); (2) every annotation contains all three rubric parts — summary, critical evaluation, and relevance; (3) the evaluation part genuinely assesses credibility/method/bias rather than only summarizing; (4) sources meet the inclusion criteria and the minimum count is met; (5) entries are ordered as specified; (6) citation data matches {{workPath}}/sources.md with nothing fabricated. An entry missing a rubric part, or with a malformed/invented citation, FAILS.\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: \"write\" })` 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": "bibliography.md"
+        }
+      ],
+      "next": "write"
+    },
+    {
+      "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/an/annotated-bibliography/versions/1.0.2/test.json

@@ -0,0 +1,105 @@
+{
+  "schemaVersion": 1,
+  "title": "Annotated bibliography from seeded source cards",
+  "objective": "Measure whether the annotated-bibliography craftbook locks style/rubric and produces uniform summary, evaluation, and relevance annotations.",
+  "tags": [
+    "data"
+  ],
+  "prompt": "In the `Annotated Bibliography Eval` project, read `source/source-cards.md` and write `bibliography.md` for a research question about rural telehealth follow-up programs. Use APA-style entries and give every source a Summary, Critical evaluation, and Relevance annotation. Do not invent sources beyond the six seeded cards.",
+  "setup": {
+    "projectName": "Annotated Bibliography Eval",
+    "about": "Self-contained eval project for annotated-bibliography. Six source cards are in workspace/source/source-cards.md.",
+    "missionObjectives": "Write a six-entry APA-style annotated bibliography with uniform three-part annotations and no invented sources.",
+    "files": [
+      {
+        "path": "source/source-cards.md",
+        "content": "# Source cards\n\nResearch question: What makes rural telehealth follow-up programs effective after hospital discharge?\n\n1. Allen, Priya. 2022. Rural Telehealth Follow-up and Readmission Risk. Journal of Rural Care. Finds video follow-up within 72 hours lowered 30-day readmissions by 11% across four clinics; limitation: observational design.\n2. Chen, Marco. 2021. Broadband Gaps in Remote Patient Monitoring. Health Systems Review. Surveys 900 households; shows connectivity failures explain many missed visits; limitation: self-reported bandwidth.\n3. Dunn, Elise. 2023. Nurse-led Coaching After Discharge. Care Transitions Quarterly. Randomized pilot, n=240; nurse coaching improved medication adherence by 18 points; limitation: single region.\n4. Garcia, Alma. 2020. Telehealth Trust in Older Adults. Community Health Methods. Qualitative interviews with 42 older adults; highlights trust, language access, and caregiver presence.\n5. Patel, Neel. 2024. Cost Model for Hybrid Follow-up. Rural Health Economics. Compares phone, video, and in-person follow-ups; finds hybrid model breaks even after 1,600 annual encounters.\n6. Williams, June. 2019. Privacy Concerns in Home Monitoring. Nursing Informatics. Reviews consent and device-sharing risks; older but still useful for privacy checklist.\n"
+      }
+    ],
+    "worker": {
+      "name": "Bente",
+      "role": "Researcher"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "workspace/bibliography.md has six source-grounded entries with APA-style citation data plus summary, credibility/method evaluation, and relevance notes.",
+    "deliverables": [
+      {
+        "path": "bibliography.md",
+        "kind": "markdown-doc",
+        "minBytes": 2600,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "bibliography.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "bibliography.md",
+            "pattern": "Allen|Chen|Dunn|Garcia|Patel|Williams",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "bibliography.md",
+            "pattern": "Summary|Critical evaluation|Relevance",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "bibliography.md",
+            "pattern": "APA|telehealth|rural|readmission",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "bibliography.md",
+            "pattern": "observational|limitation|bias|method",
+            "flags": "i"
+          },
+          {
+            "kind": "notContains",
+            "file": "bibliography.md",
+            "pattern": "Retrieved from Wikipedia|unknown author",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "bibliography.md",
+      "kind": "markdown"
+    },
+    "axes": [
+      {
+        "name": "accuracy",
+        "description": "Computed figures are correct against the seeded records."
+      },
+      {
+        "name": "grounding",
+        "description": "Every claim traces to the seeded data — nothing invented."
+      },
+      {
+        "name": "structure",
+        "description": "Tables and sections make the analysis checkable at a glance."
+      },
+      {
+        "name": "actionability",
+        "description": "Recommendations follow from the findings and are concretely doable."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "rubric uniformity",
+    "source coverage",
+    "critical evaluation"
+  ]
+}

package/data/craftbook-templates/an/annual-document-review/versions/1.0.2/craftbook.json

@@ -0,0 +1,152 @@
+{
+  "id": "annual-document-review",
+  "name": "Annual Document Review",
+  "description": "Walk a life binder's document roster once a year: inventory what is present and missing, confirm each present document is still current (right beneficiaries, right address, not expired), record the review so staleness tracking resets, and produce a prioritized to-do for the gaps. Locks the currency criteria per document kind FIRST, then verifies each record, then reports — so the binder's completeness claim is re-earned annually instead of rotting silently. Use for annual life-binder reviews, estate document checkups, 'is our paperwork still right', and insurance/beneficiary reviews.\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. Scope the review (reviewer) — inventory the roster, currency criteria per document kind → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Verify each document (reviewer) — walk the roster with the owner, record reviews, list gaps → gated on artifact `{{workPath}}/verify.md` (markdown-notes)\n3. Write the binder report (copywriter) — binder health report with a ranked to-do → gated on artifact `{{workPath}}/report.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": "scope",
+  "triggers": [
+    "annual document review",
+    "review the binder",
+    "check our documents",
+    "is our paperwork current",
+    "beneficiary review"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope the review",
+      "description": "inventory the roster, currency criteria per document kind",
+      "prompt": "Define the review before checking anything. Step 1: read documents.json from the workspace and inventory the roster — which documents are present, which are missing, and when each present one was last reviewed. Step 2: define the currency criteria per document kind — e.g. a will is current if it reflects today's family situation and executor; insurance is current if coverage amounts and beneficiaries are right; contact lists and password locations are current if verified within the year; anything expired or superseded is not current. Step 3: define what counts as done for a missing document — a concrete next step (who to call, what to draft), not just 'get one'. Step 4: write an acceptance-criteria checklist ('every roster record verified with the owner or explicitly deferred', 'every verified document gets its review recorded so staleness resets', 'every missing document has a named next step', 'the report ranks gaps by consequence, not alphabetically'). Call write_task_note with the inventory counts + criteria + checklist and write {{workPath}}/scope.md.\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": "reviewer",
+      "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": "verify"
+    },
+    {
+      "id": "verify",
+      "name": "Verify each document",
+      "description": "walk the roster with the owner, record reviews, list gaps",
+      "prompt": "Verify the roster against the locked criteria. Step 1: walk each present document with the owner — ask the kind-specific currency questions from the scope and record the answer. Step 2: for each document confirmed current, record the review through the project's binder capabilities (or note the update for documents.json) so the reviewedAt tracking resets. Step 3: for each document that fails its criteria, record what specifically is out of date. Step 4: for each missing document, confirm the next step with the owner and capture it. Step 5: write the structured results — verified, out-of-date (with why), missing (with next step) — to {{workPath}}/verify.md, and call write_task_note with the counts per bucket.\n\nThe deliverable `{{workPath}}/verify.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}}/verify.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/verify.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/verify.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "verify",
+        "maxAttempts": 3
+      },
+      "next": "report"
+    },
+    {
+      "id": "report",
+      "name": "Write the binder report",
+      "description": "binder health report with a ranked to-do",
+      "prompt": "Write the binder health report from the verification. Step 1: open with the headline — how many documents are in place and current, out of the full roster. Step 2: list what was verified current this review, one line each. Step 3: give the ranked to-do — out-of-date and missing documents ordered by consequence (what happens to the family if this one is needed and wrong), each with its concrete next step and who acts. Step 4: note the next review date one year out. Step 5: keep the tone calm and practical — the binder exists to make a hard day easier, and the report should read that way. Write {{workPath}}/report.md and call write_task_note with the report path and the to-do count.\n\nThe deliverable `{{workPath}}/report.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": "copywriter",
+      "advanceWhen": {
+        "file": "{{workPath}}/report.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/report.md",
+            "bytes": 1500,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/report.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": "Open {{workPath}}/report.md and {{workPath}}/verify.md and verify every criterion from {{workPath}}/scope.md: every roster record was verified or explicitly deferred, every verified document had its review recorded (spot-check two against the roster), every missing or out-of-date document carries a concrete named next step, and the to-do is ranked by consequence with a stated rationale. Write PASS/FAIL per criterion; on any failure, name the gap and loop back to report.\n\nThe deliverable lives in the project's artifacts drawer — open `{{workPath}}/report.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}}/report.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"
+}

package/data/craftbook-templates/an/annual-document-review/versions/1.0.2/test.json

@@ -0,0 +1,113 @@
+{
+  "schemaVersion": 1,
+  "title": "Annual Document Review smoke eval",
+  "objective": "Self-contained smoke eval for the Annual Document Review craftbook using the corpus generic harness.",
+  "tags": [
+    "corpus"
+  ],
+  "prompt": "We need a annual document review for the returns-desk work. The backgound info is in source/brief.md — have a look and write it up in tasks/eval/report.md.",
+  "setup": {
+    "projectName": "Annual Document Review Eval",
+    "about": "Self-contained eval project for annual-document-review. Seeded inputs are under workspace/source or workspace/fixtures; final deliverable is workspace/report.md.",
+    "missionObjectives": "Use the Annual Document Review craftbook/template, read the seeded local fixtures, and write tasks/eval/report.md without network calls, real credentials, or live services.",
+    "files": [
+      {
+        "path": "source/brief.md",
+        "content": "# Annual Document Review 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: annual-document-review - Annual Document Review.\n"
+      }
+    ],
+    "worker": {
+      "name": "Jules",
+      "role": "Writer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "tasks/eval/report.md is a source-grounded markdown artifact with required facts and actions.",
+    "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": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "source/brief\\.md|Brief",
+            "flags": "i",
+            "label": "cite or name the seeded source"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "18\\b[\\s\\S]{0,30}\\bSKUs",
+            "flags": "i",
+            "label": "include the 18-SKU pilot scope"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "18\\s*hours|6\\s*hours",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "14\\.2%|8\\.9%",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "status silence|photo submission",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/report.md",
+            "pattern": "automated status emails|barcode|Finance",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/report.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": [
+    "source-grounded prose",
+    "section structure",
+    "next-action clarity"
+  ]
+}