@bendyline/gilde 0.1.38 → 0.1.40

package/data/craftbook-templates/co/config-scaffold/versions/1.0.2/test.json

@@ -0,0 +1,123 @@
+{
+  "schemaVersion": 1,
+  "title": "Project package config scaffold",
+  "objective": "Measure whether the config-scaffold craftbook scopes runtime/tooling choices and writes a coherent package.json.",
+  "tags": [
+    "corpus"
+  ],
+  "prompt": "In the `Config Scaffold Eval` project, write `package.json` for a TypeScript Vite widget library called `harbor-widget-kit`. Include module type, exports, files, engines, scripts for dev/build/test/lint/typecheck, dependencies for @vitejs/plugin-react or vite/react as appropriate, devDependencies for TypeScript, Vitest, ESLint, and Prettier, and a packageManager field. Keep it valid JSON.",
+  "setup": {
+    "projectName": "Config Scaffold Eval",
+    "about": "Self-contained eval project for config-scaffold. The deliverable is workspace/package.json.",
+    "missionObjectives": "Produce a valid, coherent package.json scaffold for a TypeScript Vite widget library.",
+    "files": [],
+    "worker": {
+      "name": "Melle",
+      "role": "Developer"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "package.json is valid JSON and includes the expected package metadata, scripts, engines, and tool dependencies.",
+    "deliverables": [
+      {
+        "path": "package.json",
+        "kind": "json",
+        "minBytes": 600,
+        "checks": [
+          {
+            "kind": "sniff",
+            "file": "package.json",
+            "sniff": "json-valid"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "harbor-widget-kit",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "scripts",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "dev|build|test|lint|typecheck",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "dependencies|devDependencies",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "typescript|tsc",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "vite",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "vitest",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "eslint|prettier",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "package.json",
+            "pattern": "engines|packageManager",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "package.json",
+      "kind": "json"
+    },
+    "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": [
+    "config completeness",
+    "tooling coherence",
+    "valid JSON"
+  ]
+}

package/data/craftbook-templates/co/content-accuracy-review/manifest.json

@@ -3,6 +3,7 @@
   "kind": "craftbook-template",
   "id": "content-accuracy-review",
   "role": "maintenance-review",
+  "category": "research",
   "name": "Content Accuracy Review",
   "description": "Fact-check a piece of content for accuracy and produce a claim-by-claim verification report. First scopes the content and extracts every checkable claim (statistics, dates, quotes, attributions, causal and factual statements) and locks a verification standard, then verifies each claim against authoritative sources rating it true/false/misleading/unverifiable with the supporting evidence, then writes a report flagging every inaccuracy with the correction and source plus an overall trust rating. Use this for a fact-check, accuracy review, content verification, checking claims and statistics, source-checking an article, or vetting content before publishing — it returns a claim-level verdict backed by cited sources rather than a vague 'seems accurate'.",
   "tags": [

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

@@ -0,0 +1,153 @@
+{
+  "id": "content-accuracy-review",
+  "name": "Content Accuracy Review",
+  "description": "Fact-check a piece of content for accuracy and produce a claim-by-claim verification report. First scopes the content and extracts every checkable claim (statistics, dates, quotes, attributions, causal and factual statements) and locks a verification standard, then verifies each claim against authoritative sources rating it true/false/misleading/unverifiable with the supporting evidence, then writes a report flagging every inaccuracy with the correction and source plus an overall trust rating. Use this for a fact-check, accuracy review, content verification, checking claims and statistics, source-checking an article, or vetting content before publishing — it returns a claim-level verdict backed by cited sources rather than a vague 'seems accurate'.\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 + extract claims (reviewer) — extract every checkable claim, lock the standard → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Verify each claim (researcher) — check claims against sources, rate each → gated on artifact `{{workPath}}/verify.md` (markdown-notes)\n3. Write the accuracy report (reviewer) — claim-by-claim verdicts + corrections → gated on artifact `{{workPath}}/accuracy-review.md` (markdown-report)\n\nSource discipline:\n\n- Extract only checkable claims that appear in the supplied content.\n- Verify against the supplied or fetched sources; never create extra claims, companies, metrics, quotes, or source sections to make the report longer.\n- Preserve enough of the original claim text that a reader can trace each verdict back to the draft.\n- Mark unsupported but plausible claims as `UNVERIFIABLE` or `FALSE` according to the evidence, not as true by assumption.\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "fact check this",
+    "accuracy review",
+    "verify these claims",
+    "is this content accurate",
+    "source-check this article",
+    "vet this before publishing"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope + extract claims",
+      "description": "extract every checkable claim, lock the standard",
+      "prompt": "Step 1: Read the content and EXTRACT every checkable claim into a numbered list — statistics and numbers, dates, named quotes and their attributions, causal statements ('X causes Y'), definitions, and any assertion presented as fact. Step 2: For each claim, note what kind it is and what evidence would settle it. Step 3: Separate verifiable facts from opinion/analysis (opinions are out of scope for accuracy but note unsupported opinions stated as fact). Step 4: Write the acceptance-criteria checklist / verification standard, e.g. 'every factual claim is extracted', 'each claim is checked against at least one authoritative source', 'a claim with no reliable source is marked unverifiable, not assumed true', 'corrections cite their source'. Step 5: Note the source quality bar (primary/official sources over secondary). Write the numbered claim list + standard to `write_task_note` AND `{{workPath}}/scope.md`. No verification 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": "verify"
+    },
+    {
+      "id": "verify",
+      "name": "Verify each claim",
+      "description": "check claims against sources, rate each",
+      "prompt": "Step 1: For EACH numbered claim, find authoritative evidence (use web search/fetch if available, prioritizing primary and official sources; otherwise reason from well-established knowledge and clearly mark anything you cannot externally confirm). Step 2: Rate each claim: TRUE (supported), FALSE (contradicted), MISLEADING (technically true but framed to deceive, or missing critical context), or UNVERIFIABLE (no reliable source found). Step 3: For every claim capture: the rating, the supporting source(s) with title/URL, a one-line evidence note, and — for false/misleading — the CORRECT statement. Step 4: Be adversarial: check whether a statistic is current, whether a quote is accurate and in context, and whether an attribution is right. Step 5: Do NOT mark a claim true just because it is plausible — require evidence, and prefer 'unverifiable' to a guess. Good looks like: each claim with a verdict and a citable source. Stage the verified claim table to `write_task_note`. Do not write the report yet.\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": "researcher",
+      "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 accuracy report",
+      "description": "claim-by-claim verdicts + corrections",
+      "prompt": "Step 1: Open `{{workPath}}/accuracy-review.md` and write an overall trust rating (e.g. Accurate / Mostly accurate with fixes / Significant inaccuracies) with the count of false/misleading/unverifiable claims. Step 2: Add an '## Issues' section listing every FALSE, MISLEADING, and UNVERIFIABLE claim first: the original claim, the verdict, the correct statement, and the cited source. Step 3: Add a '## Verified claims' section confirming the claims that checked out (with sources), so the review's rigor is visible. Step 4: Add a '## Sources' list of everything consulted. Step 5: Note any claim that needs a subject-matter expert. Every issue must include a correction and a source citation; never assert a correction without evidence. On a loop-back, fix only the named gaps. `write_task_note` the report path and the counts by verdict.\n\nThe deliverable `{{workPath}}/accuracy-review.md` lands in the project's artifacts drawer — write it with `write_artifact` and read it back with `read_artifact`; the shipped workspace stays untouched.",
+      "suggestedRole": "reviewer",
+      "advanceWhen": {
+        "file": "{{workPath}}/accuracy-review.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/accuracy-review.md",
+            "bytes": 1500,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/accuracy-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}}/accuracy-review.md` and grade it against the scope standard. Check EACH criterion and write PASS/FAIL with a one-line reason: (1) every checkable claim from the content was extracted and addressed; (2) each claim has a verdict (true/false/misleading/unverifiable); (3) each false/misleading claim has a correction AND a cited source; (4) verified claims cite their sources; (5) unverifiable claims are marked as such rather than assumed true; (6) a sources list is present; (7) the overall trust rating matches the findings. If any criterion fails, name the exact gap and loop back.\n\nThe deliverable lives in the project's artifacts drawer — open `{{workPath}}/accuracy-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}}/accuracy-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/content-accuracy-review/versions/1.0.2/test.json

@@ -0,0 +1,152 @@
+{
+  "schemaVersion": 1,
+  "title": "Claim-level accuracy review from evidence pack",
+  "objective": "Measure whether the content-accuracy-review craftbook extracts checkable claims and writes sourced true/false/misleading/unverifiable verdicts.",
+  "tags": [
+    "corpus"
+  ],
+  "prompt": "In the `Content Accuracy Review Eval` project, read only `source/article.md` and `source/evidence-pack.md`, then write `tasks/eval/accuracy-review.md`. Extract the six checkable claims in the article, rate each true/false/misleading/unverifiable, cite the evidence pack, give corrections, and include an overall trust rating. Do not add claims, companies, sources, or metrics beyond the seeded article and evidence pack.",
+  "setup": {
+    "projectName": "Content Accuracy Review Eval",
+    "about": "Self-contained eval project for content-accuracy-review. The draft article and evidence pack are seeded in workspace/source.",
+    "missionObjectives": "Produce a claim-by-claim accuracy review with verdicts, corrections, sources, and a trust rating.",
+    "files": [
+      {
+        "path": "source/article.md",
+        "content": "# Draft article\n\nHarborCRM launched guided returns intake in February 2026. The pilot covered all SKUs and reduced median first response from 18 hours to 4 hours. Refund leakage fell from 14.2% to 8.9%. Support escalations dropped 21% after agents received recommended resolution scripts. The company guarantees national rollout in July.\n"
+      },
+      {
+        "path": "source/evidence-pack.md",
+        "content": "# Evidence pack\n\n- Launch memo: guided returns intake pilot launched in March 2026, not February.\n- Pilot scope: 18 home-office accessory SKUs, not all SKUs.\n- Metrics dashboard: median first response improved from 18 hours to 6 hours.\n- Metrics dashboard: refund leakage fell from 14.2% to 8.9%.\n- Support report: escalations dropped 21% after recommended resolution scripts were added.\n- Roadmap note: national rollout is proposed for July but not guaranteed.\n"
+      }
+    ],
+    "worker": {
+      "name": "Kiki",
+      "role": "Fact Checker"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "workspace/accuracy-review.md extracts seeded claims, assigns verdicts, cites evidence, corrects false/misleading claims, and gives an overall trust rating.",
+    "deliverables": [
+      {
+        "path": "tasks/eval/accuracy-review.md",
+        "kind": "markdown-report",
+        "minBytes": 1600,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "TRUE|FALSE|MISLEADING|UNVERIFIABLE|trust rating",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "February|March|all SKUs|18 home-office",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "18\\s*hours|4\\s*hours|6\\s*hours|14\\.2%|8\\.9%|21%",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "correction|correct statement|source|evidence",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "guarantees national rollout|proposed for July",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "February[\\s\\S]{0,250}False|False[\\s\\S]{0,250}February",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "all SKUs[\\s\\S]{0,250}False|False[\\s\\S]{0,250}all SKUs",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "4\\s*hours[\\s\\S]{0,250}(False|Misleading)|(False|Misleading)[\\s\\S]{0,250}4\\s*hours",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "14\\.2%[\\s\\S]{0,250}True|8\\.9%[\\s\\S]{0,250}True|True[\\s\\S]{0,250}14\\.2%",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "21%[\\s\\S]{0,250}True|True[\\s\\S]{0,250}21%",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "guarantees[\\s\\S]{0,300}(False|Misleading)|(False|Misleading)[\\s\\S]{0,300}guarantees",
+            "flags": "i"
+          },
+          {
+            "kind": "notContains",
+            "file": "tasks/eval/accuracy-review.md",
+            "pattern": "Q3|AlphaCorp|enterprise software|cash reserve|dominant market|CEO announced|RESTful API|stress tests",
+            "flags": "i",
+            "label": "do not invent review claims or evidence beyond the seeded article"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/accuracy-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": [
+    "claim extraction",
+    "evidence-backed verdicts",
+    "corrections"
+  ]
+}

package/data/craftbook-templates/co/content-deck/manifest.json

@@ -3,6 +3,7 @@
   "kind": "craftbook-template",
   "id": "content-deck",
   "role": "general",
+  "category": "design",
   "name": "Slide Deck from Content",
   "description": "Turn a body of source content into a presentation deck. Outlines the narrative first, designs a consistent slide look, then builds the deck (HTML slides), evaluating for narrative flow and visual consistency.",
   "tags": [

package/data/craftbook-templates/co/contract-template/manifest.json

@@ -3,6 +3,7 @@
   "kind": "craftbook-template",
   "id": "contract-template",
   "role": "general",
+  "category": "business",
   "name": "Contract Template (Non-Legal-Advice)",
   "description": "Draft a reusable, plain-language agreement template — service agreement, NDA, freelance contract, or terms — with clearly marked fill-in fields, structured numbered clauses, and a prominent non-legal-advice disclaimer. Scopes the parties, obligations, and risk allocation FIRST, drafts clear enforceable-style clauses with placeholders for the variable terms, then has a reviewer check completeness and flag anything that genuinely needs a lawyer. NOT legal advice — every output must say so. Use for templated agreements you will adapt and have counsel review.",
   "tags": [

package/data/craftbook-templates/co/contract-template/versions/1.0.2/craftbook.json

@@ -0,0 +1,157 @@
+{
+  "id": "contract-template",
+  "name": "Contract Template (Non-Legal-Advice)",
+  "description": "Draft a reusable, plain-language agreement template — service agreement, NDA, freelance contract, or terms — with clearly marked fill-in fields, structured numbered clauses, and a prominent non-legal-advice disclaimer. Scopes the parties, obligations, and risk allocation FIRST, drafts clear enforceable-style clauses with placeholders for the variable terms, then has a reviewer check completeness and flag anything that genuinely needs a lawyer. NOT legal advice — every output must say so. Use for templated agreements you will adapt and have counsel review.\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 agreement (planner) — parties, obligations, term, payment, risk allocation, fill-in fields → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Draft the clauses (copywriter) — clear numbered clauses with marked placeholders + disclaimer → gated on `contract.md` (markdown-doc)\n3. Review the template (reviewer) — check clause completeness, placeholders, and lawyer-needed flags → gated on artifact `{{workPath}}/review.md` (markdown-notes)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "contract template",
+    "draft an nda",
+    "service agreement template",
+    "freelance contract",
+    "terms template",
+    "agreement document"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope the agreement",
+      "description": "parties, obligations, term, payment, risk allocation, fill-in fields",
+      "prompt": "Map the deal structure before drafting clauses. Step 1: identify the agreement type and the parties and their roles. Step 2: list each party's core obligations and what each receives. Step 3: define the variable terms that must become FILL-IN FIELDS (names, dates, amounts, term length, jurisdiction) and mark them as placeholders like [PARTY_A_NAME]. Step 4: decide the risk-allocation clauses the template needs (confidentiality, IP ownership, liability limits, termination, dispute resolution, governing law) appropriate to the type. Step 5: note that this is a TEMPLATE and not legal advice — a disclaimer is mandatory in the output. Step 6: write an acceptance-criteria checklist ('all parties + roles defined', 'every variable is a marked fill-in field', 'standard protective clauses present for this type', 'termination + governing-law clauses included', 'a prominent non-legal-advice disclaimer is required'). Write the structure + field list + checklist via write_task_note AND to {{workPath}}/scope.md. No clause prose 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": "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": "draft"
+    },
+    {
+      "id": "draft",
+      "name": "Draft the clauses",
+      "description": "clear numbered clauses with marked placeholders + disclaimer",
+      "prompt": "Draft the agreement as contract.md in clear, plain language. Step 1: lead with a bold, unmissable disclaimer: this is a template for informational purposes, NOT legal advice, and should be reviewed by a qualified attorney before use. Step 2: write a preamble identifying the parties using the [PLACEHOLDER] fields and the effective date. Step 3: write the body as NUMBERED clauses (1, 1.1, 2 ...) — one obligation/topic per clause — covering scope of work/subject, payment terms, term and termination, confidentiality, IP ownership, liability/indemnity, dispute resolution, and governing law, each appropriate to the agreement type. Step 4: keep every variable as a clearly marked fill-in field so the user adapts it safely; add brief [GUIDANCE: ...] notes where a choice must be made. Step 5: end with a signature block (party, name, title, date lines). Use precise but readable language; avoid ambiguity in obligations. On a loop-back, fix ONLY the named gaps. Write the path + clause count via write_task_note.",
+      "suggestedRole": "copywriter",
+      "advanceWhen": {
+        "file": "contract.md",
+        "minBytes": 1,
+        "sniff": "nonempty"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "contract.md",
+            "bytes": 500
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkContains",
+            "scope": "standard",
+            "inputs": {
+              "file": "contract.md",
+              "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+              "flags": "i"
+            }
+          }
+        ],
+        "onReject": "draft",
+        "maxAttempts": 3
+      },
+      "next": "review"
+    },
+    {
+      "id": "review",
+      "name": "Review the template",
+      "description": "check clause completeness, placeholders, and lawyer-needed flags",
+      "prompt": "Review the template for completeness and honesty about its limits. Step 1: confirm the non-legal-advice disclaimer is present and prominent. Step 2: verify every scoped clause type is present and every variable term is a clearly marked fill-in field (no hard-coded names/amounts leaked in). Step 3: check the signature block and that termination + governing-law clauses exist. Step 4: flag any clause that genuinely requires a lawyer's judgment (e.g. liability caps, jurisdiction-specific enforceability) with a [LAWYER REVIEW] note rather than overstating certainty. Step 5: list concrete gaps with the clause number and fix, recorded as PASS/FAIL against the scope checklist. Write the findings via write_task_note AND to {{workPath}}/review.md. Gaps must be fixed in the draft phase before this passes.\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": "reviewer",
+      "advanceWhen": {
+        "file": "{{workPath}}/review.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "contract.md",
+            "bytes": 500
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkContains",
+            "scope": "standard",
+            "inputs": {
+              "file": "contract.md",
+              "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+              "flags": "i"
+            }
+          }
+        ],
+        "onReject": "review",
+        "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": "QA the contract template against the locked criteria. Step 1: read contract.md. Step 2: confirm FIRST that a prominent non-legal-advice disclaimer is present — if missing, FAIL immediately and loop back. Step 3: check EACH remaining criterion: parties + roles defined via placeholders, every variable is a clearly marked fill-in field, standard protective clauses for this type are present, termination + governing-law clauses included, clauses are numbered and unambiguous, and a signature block exists. Step 4: confirm lawyer-judgment items carry a [LAWYER REVIEW] flag rather than false certainty, folding in {{workPath}}/review.md. Write PASS/FAIL per criterion with the offending clause on any FAIL.\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: \"draft\" })` 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": "contract.md"
+        }
+      ],
+      "next": "draft"
+    },
+    {
+      "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"
+}