@bendyline/gilde 0.1.37 → 0.1.39

package/data/craftbook-templates/co/cookbook-layout/versions/1.0.2/test.json

@@ -0,0 +1,162 @@
+{
+  "schemaVersion": 1,
+  "title": "Lay Out the Cookbook — tested recipes to heirloom pages",
+  "objective": "Measure whether the cookbook-layout craftbook assembles the seeded recipe shelf into a print-ready book: every tested recipe in its chapter with its provenance line, untested recipes listed at the back, family phrasings preserved.",
+  "tags": [
+    "html-page"
+  ],
+  "prompt": "i think weve got enough tested recipes — can you lay out the cookbok? one per grandchild eventually so make it printable",
+  "setup": {
+    "projectName": "Van Daalen Family Recipe Book",
+    "about": "Family recipe book project for the Van Daalens. The recipe index is workspace/recipes.json with one markdown file per recipe under workspace/recipes; the laid-out book goes under workspace/book.",
+    "missionObjectives": "Tested recipes go in with their provenance intact; untested ones are listed for the next round, never silently dropped. Everything is grounded in the shelf — no live services, no real credentials.",
+    "files": [
+      {
+        "path": "recipes.json",
+        "content": "{\n  \"family\": \"Van Daalen\",\n  \"chapters\": [\"Soups & Starters\", \"Mains\", \"Breads\", \"Sweets\"],\n  \"recipes\": [\n    { \"id\": \"split-pea\", \"title\": \"Oma Trudy's Split Pea Soup\", \"chapter\": \"Soups & Starters\", \"provenance\": \"Oma Trudy's — came over from Groningen in 1954\", \"tested\": true, \"file\": \"recipes/split-pea.md\" },\n    { \"id\": \"brown-bread\", \"title\": \"Saturday Brown Bread\", \"chapter\": \"Breads\", \"provenance\": \"Grandpa Wim's — baked every Saturday for thirty years\", \"tested\": true, \"file\": \"recipes/brown-bread.md\" },\n    { \"id\": \"pot-pie\", \"title\": \"Aunt Willa's Chicken Pot Pie\", \"chapter\": \"Mains\", \"provenance\": \"Aunt Willa's — the funeral casserole that became a birthday request\", \"tested\": true, \"file\": \"recipes/pot-pie.md\" },\n    { \"id\": \"strawberry-cake\", \"title\": \"Midsummer Strawberry Cake\", \"chapter\": \"Sweets\", \"provenance\": \"Mom's — invented the summer the freezer broke\", \"tested\": true, \"file\": \"recipes/strawberry-cake.md\" },\n    { \"id\": \"campfire-chili\", \"title\": \"Pa's Campfire Chili\", \"chapter\": \"Mains\", \"provenance\": \"Pa's — written on the inside of a matchbook, translated by Ria\", \"tested\": true, \"file\": \"recipes/campfire-chili.md\" },\n    { \"id\": \"lemon-custard\", \"title\": \"Lemon Custard for Sick Days\", \"chapter\": \"Sweets\", \"provenance\": \"Oma Trudy's — the one that fixed every flu since 1962\", \"tested\": true, \"file\": \"recipes/lemon-custard.md\" },\n    { \"id\": \"oliebollen\", \"title\": \"The Contested Oliebollen\", \"chapter\": \"Sweets\", \"provenance\": \"disputed between Ria and Aunt Willa — two versions exist\", \"tested\": false, \"file\": \"recipes/oliebollen.md\" },\n    { \"id\": \"maple-fudge\", \"title\": \"Maple Walnut Fudge\", \"chapter\": \"Sweets\", \"provenance\": \"from a fair stall in 1998 — nobody has dared it yet\", \"tested\": false, \"file\": \"recipes/maple-fudge.md\" }\n  ]\n}\n"
+      },
+      {
+        "path": "recipes/split-pea.md",
+        "content": "# Oma Trudy's Split Pea Soup\n\nOma Trudy's — came over from Groningen in 1954.\n\n## Ingredients\n- 500 g split peas\n- 1 smoked sausage (the good one, not the budget one — Oma's margin note)\n- 2 leeks, 1 celeriac, 2 carrots\n- a ham hock\n\n## Method\nSoak the peas while you argue about whether soaking is necessary. Simmer everything except the sausage for two hours until 'a spoon stands up in it'. Add the sausage for the last twenty minutes. Better tomorrow than today.\n"
+      },
+      {
+        "path": "recipes/brown-bread.md",
+        "content": "# Saturday Brown Bread\n\nGrandpa Wim's — baked every Saturday for thirty years.\n\n## Ingredients\n- 400 g wholemeal flour, 100 g rye\n- 7 g yeast, 10 g salt, a knob of butter\n- 350 ml warm water\n\n## Method\nKnead until your arms complain, then five minutes more. Rise under the tea towel with the blue stripe (any towel works; the blue stripe is tradition). Bake at 220 until it sounds hollow when you knock on the bottom.\n"
+      },
+      {
+        "path": "recipes/pot-pie.md",
+        "content": "# Aunt Willa's Chicken Pot Pie\n\nAunt Willa's — the funeral casserole that became a birthday request.\n\n## Ingredients\n- 1 whole chicken, poached and picked\n- carrots, peas, pearl onions\n- double crust, all butter, no shortcuts\n\n## Method\nMake the gravy from the poaching broth and do not rush it. Fill, lid, crimp with a fork the way Willa did — three presses, turn, three presses. Vent with her initials if you're feeling brave. Bake until deep gold.\n"
+      },
+      {
+        "path": "recipes/strawberry-cake.md",
+        "content": "# Midsummer Strawberry Cake\n\nMom's — invented the summer the freezer broke.\n\n## Ingredients\n- 250 g flour, 200 g sugar, 3 eggs, 200 g butter\n- a full kilo of strawberries, no apologies\n- whipped cream, double what seems sensible\n\n## Method\nBake the sponge, split it while warm. Sugar the berries and let them bleed a little. Assemble taller than is structurally wise. Eat outside or it doesn't count.\n"
+      },
+      {
+        "path": "recipes/campfire-chili.md",
+        "content": "# Pa's Campfire Chili\n\nPa's — written on the inside of a matchbook, translated by Ria.\n\n## Ingredients\n- 1 kg beef chuck, cut rough\n- two onions, 'more garlic than the recipe says'\n- dried anchos, a bottle of dark beer, a square of dark chocolate\n\n## Method\nBrown the beef in batches and don't crowd the pot — the matchbook underlines this twice. Low fire, three hours, stir when you think of it. The chocolate goes in at the end and you tell nobody.\n"
+      },
+      {
+        "path": "recipes/lemon-custard.md",
+        "content": "# Lemon Custard for Sick Days\n\nOma Trudy's — the one that fixed every flu since 1962.\n\n## Ingredients\n- 500 ml whole milk\n- 3 yolks, 80 g sugar, 30 g cornstarch\n- zest of two lemons, juice of one\n\n## Method\nWarm the milk with the zest. Whisk yolks, sugar, and cornstarch, then marry them slowly so nothing scrambles. Thicken on low until it coats the spoon. Serve warm, in the yellow bowl if the patient is under twelve.\n"
+      },
+      {
+        "path": "recipes/oliebollen.md",
+        "content": "# The Contested Oliebollen\n\nDisputed between Ria and Aunt Willa — two versions exist and neither yields.\n\n## Status\nNot yet tested side by side. The New Year's bake-off will settle it. Both versions are filed with the arbitration notes.\n"
+      },
+      {
+        "path": "recipes/maple-fudge.md",
+        "content": "# Maple Walnut Fudge\n\nFrom a fair stall in 1998 — nobody has dared it yet.\n\n## Status\nUntested. The card says 'soft-ball stage' with no thermometer number, which is why it has waited twenty-eight years.\n"
+      }
+    ],
+    "worker": {
+      "name": "Ria",
+      "role": "Designer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "book/cookbook.html is a print-ready book: all six tested recipes in their chapters with provenance lines, the two untested ones listed at the back, family phrasings preserved.",
+    "deliverables": [
+      {
+        "path": "book/cookbook.html",
+        "kind": "html-page",
+        "minBytes": 8000,
+        "checks": [
+          {
+            "kind": "cssMinBytes",
+            "bytes": 650,
+            "file": "book/cookbook.html"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "@page|@media\\s+print|page-break|break-after",
+            "flags": "i",
+            "label": "print CSS"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Split Pea",
+            "label": "soup chapter recipe present"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Brown Bread",
+            "label": "bread recipe present"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Pot Pie",
+            "label": "mains recipe present"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Strawberry Cake",
+            "label": "sweets recipe present"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Campfire Chili",
+            "label": "second mains recipe present"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Lemon Custard",
+            "label": "second sweets recipe present"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "Groningen|Oma Trudy",
+            "label": "provenance preserved"
+          },
+          {
+            "kind": "contains",
+            "file": "book/cookbook.html",
+            "pattern": "still to test|to be tested|untested|not yet tested",
+            "flags": "i",
+            "label": "untested recipes listed, not dropped"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "book/cookbook.html",
+      "kind": "html"
+    },
+    "axes": [
+      {
+        "name": "heirloom feel",
+        "description": "The book reads like a family object — chapter intros in the family's voice, not food-blog prose."
+      },
+      {
+        "name": "completeness",
+        "description": "All six tested recipes appear in their chapters; the two untested ones are listed at the back."
+      },
+      {
+        "name": "typography",
+        "description": "Type set for reading at arm's length in a working kitchen; one recipe per printed page."
+      },
+      {
+        "name": "provenance preservation",
+        "description": "Every recipe keeps its provenance line and the family's own phrasings survive in the methods."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "inclusion rule honored",
+    "provenance intact",
+    "print-ready book"
+  ]
+}

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

@@ -0,0 +1,153 @@
+{
+  "id": "copy-review",
+  "name": "Editorial & Tone Review",
+  "description": "Run an editorial review on a piece of content for clarity, correctness, and brand voice, then produce a line-edit report and a clean revised version. First scopes the audience, purpose, and voice/style rules and locks an editorial checklist (grammar and spelling, clarity and concision, consistent terminology, active voice, reading level, tone match, scannability, and CTA strength), then reviews the copy line-by-line against those rules marking every issue with a suggested rewrite, then writes a report of categorized edits plus a clean revised draft. Use this for a copy review, editorial pass, proofreading, line edit, tone/voice review, or tightening writing — it returns specific tracked edits with rewrites and a polished final draft rather than 'tighten this up'.\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 editorial review (reviewer) — lock audience, purpose, voice, and an edit checklist → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Line-edit the copy (copywriter) — mark every issue with a concrete rewrite → gated on artifact `{{workPath}}/review.md` (markdown-notes)\n3. Write the edit report + clean draft (copywriter) — categorized edits plus a revised draft → gated on artifact `{{workPath}}/copy-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": "scope",
+  "triggers": [
+    "copy review",
+    "editorial review",
+    "proofread this",
+    "line edit this",
+    "review the tone",
+    "tighten this writing"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope the editorial review",
+      "description": "lock audience, purpose, voice, and an edit checklist",
+      "prompt": "Step 1: Read the content and identify the audience, the purpose (inform / persuade / convert / instruct), and the desired voice/tone (e.g. friendly-expert, formal, playful) — infer from any style guide or sample if none is given. Step 2: Write an editorial acceptance-criteria checklist covering: grammar/spelling/punctuation correctness, clarity (no ambiguous or convoluted sentences), concision (cut filler, redundancy, hedging), consistent terminology and capitalization, active voice and strong verbs, an appropriate reading level for the audience, tone matching the target voice, scannability (headings, short paragraphs, lists where helpful), and CTA/heading strength where applicable. Step 3: For each item state the pass condition for THIS piece (e.g. 'reading level ~grade 8', 'product name always rendered as <X>'). Step 4: Note the target word-count direction if length matters. Write the audience/purpose/voice + checklist to `write_task_note` AND `{{workPath}}/scope.md`. No edits 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": "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": "review"
+    },
+    {
+      "id": "review",
+      "name": "Line-edit the copy",
+      "description": "mark every issue with a concrete rewrite",
+      "prompt": "Step 1: Read the copy line-by-line. Step 2: For EACH checklist dimension, flag specific issues quoting the offending text and giving a concrete REWRITE, not just a note — e.g. 'ORIGINAL: \"We are committed to leveraging synergies\" → SUGGESTED: \"We connect your tools so they work together\"'. Step 3: Categorize each edit: Correctness (grammar/spelling/fact), Clarity, Concision, Consistency, Voice/Tone, or Structure/Scannability. Step 4: Check global consistency: terminology, capitalization, oxford-comma and number style, heading parallelism. Step 5: Assess the overall tone against the target and note where it drifts. Step 6: Track the substantive vs trivial edits so the report can prioritize. Good looks like: every flagged line has a ready-to-paste rewrite. Stage the edit list to `write_task_note`. Do not write the final report/draft yet.\n\nThe deliverable `{{workPath}}/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": "copywriter",
+      "advanceWhen": {
+        "file": "{{workPath}}/review.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/review.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/review.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "review",
+        "maxAttempts": 3
+      },
+      "next": "report"
+    },
+    {
+      "id": "report",
+      "name": "Write the edit report + clean draft",
+      "description": "categorized edits plus a revised draft",
+      "prompt": "Step 1: Open `{{workPath}}/copy-review.md` and write a 2-3 sentence editorial summary: overall clarity/voice verdict and the top theme to fix. Step 2: Add an '## Edits' section grouped by category (Correctness, Clarity, Concision, Consistency, Voice, Structure); each entry shows the ORIGINAL and the SUGGESTED rewrite. Step 3: Add a '## Revised draft' section containing the FULL clean copy with every accepted edit applied — this is the shippable version. Step 4: Note any factual claims you could not verify. The revised draft must read smoothly end-to-end and match the target voice; every edit must include a concrete rewrite — no 'reword this'. On a loop-back, fix only the named gaps without losing good edits. `write_task_note` the report path and the count of edits by category.\n\nThe deliverable `{{workPath}}/copy-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": "copywriter",
+      "advanceWhen": {
+        "file": "{{workPath}}/copy-review.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/copy-review.md",
+            "bytes": 1500,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/copy-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": "Open `{{workPath}}/copy-review.md` and grade it against the scope checklist. Check EACH criterion and write PASS/FAIL with a one-line reason: (1) every editorial dimension was assessed; (2) each edit includes the original AND a concrete rewrite; (3) edits are categorized; (4) a full revised draft is present and reads cleanly end-to-end; (5) the revised draft matches the target voice and reading level; (6) terminology/consistency was checked globally; (7) no vague 'tighten this' filler. Read the revised draft straight through to confirm it is shippable. If any criterion fails, name the exact gap and loop back.\n\nThe deliverable lives in the project's artifacts drawer — open `{{workPath}}/copy-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}}/copy-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"
+}

package/data/craftbook-templates/co/copy-review/versions/1.0.2/test.json

@@ -0,0 +1,105 @@
+{
+  "schemaVersion": 1,
+  "title": "Line-edit report and clean revised copy",
+  "objective": "Measure whether the copy-review craftbook scopes voice rules, identifies line-level issues, and produces both an edit report and revised draft.",
+  "tags": [
+    "corpus"
+  ],
+  "prompt": "In the `Copy Review Eval` project, review `source/signup-copy.md` for a calm professional operations SaaS voice. Write `tasks/eval/copy-review.md` with categorized issues, concrete rewrites, and a clean revised draft. The final draft must keep the same product facts but remove hype, vague claims, and confusing CTA language.",
+  "setup": {
+    "projectName": "Copy Review Eval",
+    "about": "Self-contained eval project for the copy-review craftbook. Source copy is in workspace/source/signup-copy.md.",
+    "missionObjectives": "Produce a line-edit report plus clean revised copy, preserving facts while improving clarity, tone, CTA, and specificity.",
+    "files": [
+      {
+        "path": "source/signup-copy.md",
+        "content": "# Signup Copy Draft\n\nTransform your support chaos instantly with the most powerful queue magic on earth. Our dashboard kind of helps managers see stuff faster and maybe make better decisions. Click here to experience effortless operational enlightenment. Teams can track ticket volume, SLA breaches, reopen rate, and carrier scan delays. The pilot setup takes two weeks and uses existing ticket exports.\n"
+      }
+    ],
+    "worker": {
+      "name": "Sofie",
+      "role": "Writer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "workspace/copy-review.md contains categorized issues, specific rewrites, and a clean revised draft preserving seeded facts.",
+    "deliverables": [
+      {
+        "path": "tasks/eval/copy-review.md",
+        "kind": "markdown-report",
+        "minBytes": 1700,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/copy-review.md",
+            "pattern": "Issues|Findings|Line edits",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/copy-review.md",
+            "pattern": "Rewrite|Suggested rewrite|Before|After",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/copy-review.md",
+            "pattern": "Clean revised draft|Revised copy",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/copy-review.md",
+            "pattern": "ticket volume|SLA breaches|reopen rate|carrier scan delays",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/copy-review.md",
+            "pattern": "two weeks|existing ticket exports",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/copy-review.md",
+            "pattern": "calm|professional|specific|CTA|voice",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/copy-review.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": [
+    "line edits",
+    "voice/tone scope",
+    "clean revised draft"
+  ]
+}

package/data/craftbook-templates/co/corpus-synthesis/versions/1.0.2/craftbook.json

@@ -0,0 +1,158 @@
+{
+  "id": "corpus-synthesis",
+  "name": "Corpus Synthesis",
+  "description": "Read a provided corpus of documents and synthesize it into a single coherent overview: a planner sets the synthesis question and the extraction schema, a researcher reads every document and extracts structured notes keyed back to each source, and a copywriter synthesizes the extracts into a unified narrative organized by idea rather than by document. The ordering matters because synthesizing before reading everything privileges the first few documents, and a write-up keyed to documents instead of ideas just summarizes each file; locking a synthesis question and per-document extraction first guarantees full coverage and a real cross-document synthesis. Use this for synthesizing a folder of docs/notes/papers, 'pull together everything we know about X', cross-document summaries, knowledge consolidation, and meeting/research-note synthesis.\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 synthesis (planner) — synthesis question + per-document extraction schema → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Read & extract (researcher) — structured extract per document, keyed to the source → gated on artifact `{{workPath}}/extracts.md` (markdown-notes)\n3. Synthesize (copywriter) — unified narrative organized by idea, citing sources → gated on artifact `{{workPath}}/synthesis.md` (markdown-report)\n\nSource discipline:\n\n- Synthesize only claims that appear in the supplied corpus or extraction notes.\n- Preserve source ids such as `[1]`, `[2]`, `[3]`, and `[4]` on synthesized points so readers can trace them.\n- Do not invent systems, timelines, metrics, risks, or implementation details to make the report longer.\n- When the corpus leaves a gap, name it as an open question instead of filling it with plausible detail.\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "synthesize these documents",
+    "pull together everything on",
+    "summarize this corpus",
+    "consolidate these notes",
+    "cross-document summary"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope the synthesis",
+      "description": "synthesis question + per-document extraction schema",
+      "prompt": "1. List the corpus (every document/file in scope) and state the synthesis question in one sentence — what unified understanding the output must deliver. 2. Define a per-document extraction schema: for each doc capture source id, key claims/findings, relevant data points, and any stance or contradiction. 3. Decide the output structure: organize by theme/idea (3-6 themes), not by document, plus an open-questions/contradictions section. 4. Write an acceptance-criteria checklist (e.g. 'every document in the corpus is read and extracted', 'output is organized by theme not by file', 'each synthesized point cites the document(s) it came from', 'contradictions across documents are surfaced', 'nothing is asserted that is not in the corpus'). 5. Record decisions. Call write_task_note with the corpus list, schema, themes, 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": "read"
+    },
+    {
+      "id": "read",
+      "name": "Read & extract",
+      "description": "structured extract per document, keyed to the source",
+      "prompt": "1. Open and read EVERY document in the corpus — do not skip or sample. 2. For each, append an extract block to {{workPath}}/extracts.md following the schema: source id [n], its key claims/findings, supporting data points (with the figure), and any stance or point where it conflicts with another document. 3. Tag each extract with which output theme(s) it speaks to. 4. As you read, note explicit cross-document agreements and contradictions. 5. Stay faithful to the source — capture what the document says, not your opinion, and quote load-bearing passages. Good looks like an extract file complete enough that the writer never reopens the originals. Do not write the synthesis yet.\n\nThe deliverable `{{workPath}}/extracts.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}}/extracts.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/extracts.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/extracts.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "read",
+        "maxAttempts": 3
+      },
+      "next": "synthesize"
+    },
+    {
+      "id": "synthesize",
+      "name": "Synthesize",
+      "description": "unified narrative organized by idea, citing sources",
+      "prompt": "1. Write {{workPath}}/synthesis.md organized by the locked themes (one section each), weaving together what the documents collectively say. 2. Each synthesized point cites the document(s) [n] it draws from so a reader can trace it. 3. Open with a 1-paragraph answer to the synthesis question, then the thematic body, then an 'Open questions & contradictions' section that surfaces where the corpus disagrees or leaves gaps. 4. Synthesize across documents — do NOT produce a per-file summary list. 5. Assert only what the corpus supports (from {{workPath}}/extracts.md). 6. Add a source key so every [n] maps to a document. 7. On a loop-back, fix only the named gaps. 8. Call write_task_note with the synthesis path and which criteria now pass.\n\nBefore working, open the artifact input `{{workPath}}/extracts.md` with `read_artifact`; do not look for it in the workspace.\n\nThe deliverable `{{workPath}}/synthesis.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",
+      "consumes": [
+        {
+          "file": "{{workPath}}/extracts.md",
+          "artifact": true
+        }
+      ],
+      "advanceWhen": {
+        "file": "{{workPath}}/synthesis.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/synthesis.md",
+            "bytes": 1500,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/synthesis.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i",
+            "label": "at least one markdown heading",
+            "artifact": true
+          }
+        ],
+        "onReject": "synthesize",
+        "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}}/synthesis.md and check EACH criterion PASS/FAIL: (1) every document in {{workPath}}/scope.md appears in {{workPath}}/extracts.md (full coverage) and is cited somewhere in the synthesis; (2) the output is organized by theme, not file-by-file; (3) each synthesized point cites its source document(s); (4) an open-questions/contradictions section surfaces real cross-document disagreement; (5) spot-check 3 points against {{workPath}}/extracts.md to confirm nothing was asserted beyond the corpus. A per-file summary list, or a synthesis missing documents, FAILS.\n\nThe deliverable lives in the project's artifacts drawer — open `{{workPath}}/synthesis.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: \"synthesize\" })` 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}}/synthesis.md",
+          "artifact": true
+        }
+      ],
+      "next": "synthesize"
+    },
+    {
+      "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"
+}