@bendyline/gilde 0.1.51 → 0.1.53

package/data/craftbook-templates/pu/pull-request-review/versions/1.9.1/test.json

@@ -0,0 +1,141 @@
+{
+  "schemaVersion": 1,
+  "title": "Complete-corpus pull request review",
+  "objective": "Verify that Pull Request Review fans a chunk-readable PR corpus out to per-batch reviewers, deterministically merges their exact coverage shards into a provenance-checked run-wide ledger, verifies an API use against a late definition held by a different batch, finds a late-file defect, reports CI evidence separately, and writes a cited local verdict without modifying source or posting to GitHub.",
+  "tags": [
+    "external",
+    "pull-request",
+    "large-corpus",
+    "coverage",
+    "fanout"
+  ],
+  "prompt": "Review the deterministic large pull-request corpus completely. Publish the review batches, let the crew review them in parallel, let the runtime merge and provenance-check their coverage shards, verify the early API call against its late definition, find the late-file authorization defect, and write tasks/eval/pr-review.md. Do not modify source or post a GitHub comment.",
+  "setup": {
+    "projectName": "Large PR Review Eval",
+    "about": "A deterministic PR-review fixture. The registered hand-authored scenario materializes more than 100 changed-file records and more than 80 KB of patch text in the artifacts corpus.",
+    "missionObjectives": "Account for every changed path, avoid a false missing-API finding, catch the late authorization defect, and deliver a local request-changes report with exact path:line evidence.",
+    "files": [
+      {
+        "path": "src/api/use-assured-api.ts",
+        "content": "import { assuredApi } from './assured-api.js';\n\nexport function loadWidget(id: string): string {\n  return assuredApi(id);\n}\n"
+      },
+      {
+        "path": "src/api/assured-api.ts",
+        "content": "export function assuredApi(id: string): string {\n  return `widget:${id}`;\n}\n"
+      },
+      {
+        "path": "src/security/late-authorization.ts",
+        "content": "export function canDeleteProject(userId: string, ownerId: string): boolean {\n  if (!userId) return false;\n  // Regression: any authenticated user can delete another owner's project.\n  return true;\n}\n"
+      }
+    ],
+    "worker": {
+      "name": "Rina",
+      "role": "Reviewer",
+      "description": "A careful code reviewer who works through large change sets in bounded batches and verifies cross-file claims."
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "The batches are published, every exact batch shard is deterministically merged into a provenance-stamped 120-file ledger, the report recognizes that assuredApi exists, cites the late authorization defect, requests changes, and leaves the seeded source unchanged.",
+    "deliverables": [
+      {
+        "path": "tasks/eval/pr-review.md",
+        "kind": "markdown-report",
+        "minBytes": 500,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/pr-review.md",
+            "pattern": "Coverage:\\s*120\\s*/\\s*120\\s+changed files",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/pr-review.md",
+            "pattern": "src/security/late-authorization\\.ts(?::|\\s*\\|\\s*)4",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/pr-review.md",
+            "pattern": "Verdict:\\s*request-changes",
+            "flags": "i"
+          },
+          {
+            "kind": "notContains",
+            "file": "tasks/eval/pr-review.md",
+            "pattern": "assuredApi.{0,80}(missing|does not exist|undefined)|(?:missing|does not exist|undefined).{0,80}assuredApi",
+            "flags": "is"
+          }
+        ]
+      },
+      {
+        "path": "tasks/eval/pr-review-coverage.json",
+        "kind": "json",
+        "minBytes": 500
+      },
+      {
+        "path": "tasks/eval/pr-review/batches.json",
+        "kind": "json",
+        "minBytes": 500,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/pr-review/batches.json",
+            "pattern": "\"batchNumber\"[\\s\\S]*\"paths\"[\\s\\S]*\"records\""
+          }
+        ]
+      }
+    ],
+    "unchangedFixtures": [
+      "src/api/use-assured-api.ts",
+      "src/api/assured-api.ts",
+      "src/security/late-authorization.ts"
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/pr-review.md",
+      "kind": "markdown"
+    },
+    "axes": [
+      {
+        "name": "coverage",
+        "description": "The review demonstrates complete 120-file coverage across every batch rather than reasoning from an early prefix or from whichever shards happened to land first."
+      },
+      {
+        "name": "cross-file-verification",
+        "description": "The report does not allege that assuredApi is missing. Its definition sits in a different batch than its use, so a batch reviewer must verify against the checkout rather than against the records it was handed."
+      },
+      {
+        "name": "late-defect",
+        "description": "The report finds and cites the authorization defect deliberately placed in the final changed file — the last batch is reviewed as carefully as the first."
+      },
+      {
+        "name": "synthesis",
+        "description": "The final report draws on every per-batch observations shard, merging duplicate findings raised by different reviewers instead of dropping or double-listing them."
+      },
+      {
+        "name": "side-effects",
+        "description": "The review remains local and source fixtures stay unchanged."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "complete large-PR coverage",
+    "per-batch fanout and deterministic provenance-checked shard merging",
+    "cross-file API verification",
+    "late-file defect detection",
+    "safe local-only delivery"
+  ],
+  "extensions": {
+    "legacySimulators": [
+      {
+        "id": "large-pr-artifact-corpus",
+        "kind": "data-source",
+        "status": "implemented",
+        "description": "The hand-authored eval scenario seeds 120 per-file PR records totaling more than 80 KB, with an API use early, its valid definition late, and a real authorization defect in the final record."
+      }
+    ]
+  }
+}

package/data/craftbook-templates/pu/puzzle-pack-layout/versions/1.0.3/craftbook.json

@@ -0,0 +1,233 @@
+{
+  "id": "puzzle-pack-layout",
+  "name": "Lay Out the Puzzle Pack",
+  "description": "Bind the month's puzzles into one print-ready pack: a cover, a page per puzzle, and the solutions gathered in the back where they belong — the thing that lands on the kitchen table and disappears into pencil marks. Locks the shelf inventory and the page rules FIRST, then lays out the pack, then proofs that every puzzle made it in and no solution leaks next to its puzzle. Use for laying out the puzzle pack, binding the month's puzzles, a printable puzzle booklet, and 'put the puzzles into one pack'.\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. Inventory the shelf (puzzle-maker) — which puzzles are in, pack order, page rules → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Lay out the pack (designer) — cover, page per puzzle, solutions in the back → gated on `pack/pack.html` (html-page)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "lay out the puzzle pack",
+    "bind the puzzles into a pack",
+    "printable puzzle booklet",
+    "puzzle pack for the month",
+    "make the puzzle pack"
+  ],
+  "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}}"
+      }
+    }
+  },
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Inventory the shelf",
+      "description": "which puzzles are in, pack order, page rules",
+      "prompt": "Inventory before laying anything out. Step 1: read the puzzles/ shelf from the workspace — every puzzle's data and rendered page, its title, and when it was made. Step 2: confirm with the owner which puzzles belong in this pack and the order they should appear. Step 3: lock the page rules — a cover with the pack's title and contents, one puzzle per page with a page break between, solutions gathered as their own section at the back and never beside their puzzle, print styles throughout. Step 4: write an acceptance-criteria checklist ('every chosen shelf puzzle appears in the pack under its own title', 'one puzzle per page with clean breaks', 'all solutions in the back section, none interleaved', 'the cover lists the contents', 'the pack prints cleanly'). Call write_task_note with the puzzle count + order + 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": "puzzle-maker",
+      "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": "layout",
+      "toolPolicy": {
+        "disallowBuiltinToolsets": [
+          "ai-apps",
+          "archives",
+          "audio",
+          "browser-automation",
+          "code-execution",
+          "craftbooks",
+          "data-tables",
+          "entity-intel",
+          "git",
+          "image-intel",
+          "images",
+          "role-delegation",
+          "role-delegation-escalation",
+          "security-intel",
+          "team-management",
+          "videos",
+          "web",
+          "workspace-fs-write"
+        ],
+        "outputMedium": "artifact",
+        "additionalOutputMedia": [
+          "task-note"
+        ]
+      }
+    },
+    {
+      "id": "layout",
+      "name": "Lay out the pack",
+      "description": "cover, page per puzzle, solutions in the back",
+      "prompt": "Lay out the pack as one print-ready HTML file at pack/pack.html. Step 1: the cover — the pack's title, the month, and the contents list. Step 2: one page per puzzle, re-laid from the shelf's puzzle data — the blank grid and clues set properly for paper, never a screenshot — each starting on a fresh page. Step 3: the solutions section last, one clearly-labeled solution per puzzle, separated from the puzzles by a page break so nobody's eye falls on an answer early. Step 4: print styles throughout — an @page rule and page-break rules so every puzzle starts fresh and the grid squares stay pencil-sized. Call write_task_note with the pack path and the page count.",
+      "suggestedRole": "designer",
+      "advanceWhen": {
+        "file": "pack/pack.html",
+        "minBytes": 1,
+        "sniff": "html-complete"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "pack/pack.html",
+            "bytes": 3000
+          },
+          {
+            "kind": "htmlLint",
+            "file": "pack/pack.html"
+          },
+          {
+            "kind": "cssMinBytes",
+            "bytes": 650,
+            "file": "pack/pack.html"
+          },
+          {
+            "kind": "contains",
+            "file": "pack/pack.html",
+            "pattern": "@page|page-break|break-after|break-before",
+            "flags": "i",
+            "label": "page-break CSS"
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkHtmlComplete",
+            "scope": "standard",
+            "inputs": {
+              "file": "pack/pack.html"
+            }
+          }
+        ],
+        "onReject": "layout",
+        "maxAttempts": 4
+      },
+      "next": "evaluate",
+      "toolPolicy": {
+        "disallowBuiltinToolsets": [
+          "ai-apps",
+          "archives",
+          "artifacts",
+          "audio",
+          "browser-automation",
+          "code-execution",
+          "craftbooks",
+          "data-tables",
+          "entity-intel",
+          "git",
+          "image-intel",
+          "images",
+          "role-delegation",
+          "role-delegation-escalation",
+          "security-intel",
+          "team-management",
+          "videos",
+          "web"
+        ],
+        "outputMedium": "workspace",
+        "additionalOutputMedia": [
+          "task-note"
+        ]
+      }
+    },
+    {
+      "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 pack/pack.html against the puzzles/ shelf and verify every criterion from {{workPath}}/scope.md: every chosen puzzle appears in the pack under its own title — check the shelf listing, a missing puzzle is the classic failure — each puzzle gets its own page with a clean break, every solution sits in the back section and none beside its puzzle, the cover lists the contents, and the page-break styling would print cleanly. Write PASS/FAIL per criterion; on any failure, name the puzzle and the gap and loop back to layout.\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: \"layout\" })` 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": "pack/pack.html"
+        }
+      ],
+      "next": "layout",
+      "toolPolicy": {
+        "disallowBuiltinToolsets": [
+          "ai-apps",
+          "archives",
+          "artifacts",
+          "audio",
+          "browser-automation",
+          "code-execution",
+          "craftbooks",
+          "data-tables",
+          "entity-intel",
+          "git",
+          "image-intel",
+          "images",
+          "role-delegation",
+          "role-delegation-escalation",
+          "security-intel",
+          "team-management",
+          "videos",
+          "web",
+          "workspace-fs-write"
+        ],
+        "outputMedium": "task-note"
+      }
+    },
+    {
+      "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,
+      "toolPolicy": {
+        "disallowBuiltinToolsets": [
+          "ai-apps",
+          "archives",
+          "artifacts",
+          "audio",
+          "browser-automation",
+          "code-execution",
+          "craftbooks",
+          "data-tables",
+          "entity-intel",
+          "git",
+          "image-intel",
+          "images",
+          "role-delegation",
+          "role-delegation-escalation",
+          "security-intel",
+          "team-management",
+          "videos",
+          "web",
+          "workspace-fs-write"
+        ],
+        "outputMedium": "task-note"
+      }
+    }
+  ],
+  "version": "1.0.3",
+  "releasedAt": "2026-09-01T20:00:00Z",
+  "minGezelVersion": "1.26233"
+}

package/data/craftbook-templates/pu/puzzle-pack-layout/versions/1.0.3/test.json

@@ -0,0 +1,123 @@
+{
+  "schemaVersion": 1,
+  "title": "Lay Out the Puzzle Pack — shelf to printable booklet",
+  "objective": "Measure whether the puzzle-pack-layout craftbook binds the seeded puzzles/ shelf into one print-ready pack with a cover, a page per puzzle, and the solutions gathered at the back.",
+  "tags": [
+    "html-page"
+  ],
+  "prompt": "can you put this months three puzzles into one printable pack? solutions at the back, seperate pages",
+  "setup": {
+    "projectName": "Loon Lake Puzzle Almanac",
+    "about": "Puzzle almanac project for the Halloran family. Finished puzzles live on the workspace/puzzles shelf as data plus a rendered page each; packs go under workspace/pack.",
+    "missionObjectives": "The pack must include every chosen shelf puzzle, one per page, with solutions sectioned at the back. Everything is grounded in the shelf files — no live services, no real credentials.",
+    "files": [
+      {
+        "path": "puzzles/loon-lake-crossword.json",
+        "content": "{\n  \"title\": \"Loon Lake Crossword\",\n  \"kind\": \"crossword\",\n  \"size\": { \"rows\": 5, \"cols\": 5 },\n  \"grid\": [\"CANOE\", \"A....\", \"BAIT.\", \"I....\", \"NORA.\"],\n  \"entries\": [\n    { \"number\": 1, \"answer\": \"CANOE\", \"row\": 0, \"col\": 0, \"direction\": \"across\", \"clue\": \"What tipped over in 2019, twice\" },\n    { \"number\": 1, \"answer\": \"CABIN\", \"row\": 0, \"col\": 0, \"direction\": \"down\", \"clue\": \"Where the summer happens\" },\n    { \"number\": 2, \"answer\": \"BAIT\", \"row\": 2, \"col\": 0, \"direction\": \"across\", \"clue\": \"What Grandpa forgets to buy\" },\n    { \"number\": 3, \"answer\": \"NORA\", \"row\": 4, \"col\": 0, \"direction\": \"across\", \"clue\": \"Reigning flashlight-tag champion\" }\n  ],\n  \"solution\": \"grid as printed above\"\n}\n"
+      },
+      {
+        "path": "puzzles/loon-lake-crossword.html",
+        "content": "<!doctype html>\n<html><head><meta charset=\"utf-8\"><title>Loon Lake Crossword</title><style>td{width:2em;height:2em;border:1px solid #333;text-align:center}</style></head>\n<body><h1>Loon Lake Crossword</h1><table></table><p>Across: 1. What tipped over in 2019, twice; 2. What Grandpa forgets to buy; 3. Reigning flashlight-tag champion. Down: 1. Where the summer happens.</p></body></html>\n"
+      },
+      {
+        "path": "puzzles/garden-birds-word-search.json",
+        "content": "{\n  \"title\": \"Garden Birds Word Search\",\n  \"kind\": \"word-search\",\n  \"size\": { \"rows\": 8, \"cols\": 8 },\n  \"grid\": [\"WRENXQZC\", \"WARBLERH\", \"FINCHAAI\", \"XNUTHATC\", \"ROBINPZK\", \"JAYQWERA\", \"SISKINQD\", \"CHICKADE\"],\n  \"words\": [\"WREN\", \"WARBLER\", \"FINCH\", \"ROBIN\", \"JAY\", \"SISKIN\"],\n  \"solution\": \"words run left to right on their rows\"\n}\n"
+      },
+      {
+        "path": "puzzles/garden-birds-word-search.html",
+        "content": "<!doctype html>\n<html><head><meta charset=\"utf-8\"><title>Garden Birds Word Search</title><style>td{width:1.6em;height:1.6em;text-align:center;font-family:monospace}</style></head>\n<body><h1>Garden Birds Word Search</h1><table></table><p>Find: WREN, WARBLER, FINCH, ROBIN, JAY, SISKIN.</p></body></html>\n"
+      },
+      {
+        "path": "puzzles/porch-sudoku.json",
+        "content": "{\n  \"title\": \"Porch Sudoku\",\n  \"kind\": \"sudoku-4x4\",\n  \"size\": { \"rows\": 4, \"cols\": 4 },\n  \"grid\": [\"1..4\", \"..2.\", \".3..\", \"4..1\"],\n  \"solution\": [\"1234\", \"3421\", \"2313\", \"4132\"],\n  \"note\": \"gentle one for the porch after dinner\"\n}\n"
+      },
+      {
+        "path": "puzzles/porch-sudoku.html",
+        "content": "<!doctype html>\n<html><head><meta charset=\"utf-8\"><title>Porch Sudoku</title><style>td{width:2em;height:2em;border:1px solid #333;text-align:center}</style></head>\n<body><h1>Porch Sudoku</h1><table></table><p>Fill each row, column, and box with 1 through 4.</p></body></html>\n"
+      }
+    ],
+    "worker": {
+      "name": "Marnix",
+      "role": "Designer"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "pack/pack.html is a print-ready booklet: cover with contents, one page per shelf puzzle, and all solutions gathered in a back section.",
+    "deliverables": [
+      {
+        "path": "pack/pack.html",
+        "kind": "html-page",
+        "minBytes": 3000,
+        "checks": [
+          {
+            "kind": "cssMinBytes",
+            "bytes": 650,
+            "file": "pack/pack.html"
+          },
+          {
+            "kind": "contains",
+            "file": "pack/pack.html",
+            "pattern": "@page|page-break|break-after|break-before",
+            "flags": "i",
+            "label": "page-break CSS"
+          },
+          {
+            "kind": "contains",
+            "file": "pack/pack.html",
+            "pattern": "Loon Lake",
+            "label": "crossword included"
+          },
+          {
+            "kind": "contains",
+            "file": "pack/pack.html",
+            "pattern": "Garden Birds",
+            "label": "word search included"
+          },
+          {
+            "kind": "contains",
+            "file": "pack/pack.html",
+            "pattern": "Porch Sudoku|Sudoku",
+            "label": "sudoku included"
+          },
+          {
+            "kind": "contains",
+            "file": "pack/pack.html",
+            "pattern": "solutions?",
+            "flags": "i",
+            "label": "solutions section present"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "pack/pack.html",
+      "kind": "html"
+    },
+    "axes": [
+      {
+        "name": "layout quality",
+        "description": "Grids and clues are re-laid properly for paper — pencil-sized squares, readable clue lists, a real cover."
+      },
+      {
+        "name": "completeness",
+        "description": "All three shelf puzzles appear under their own titles; none dropped or duplicated."
+      },
+      {
+        "name": "print readiness",
+        "description": "Page breaks put one puzzle per page and keep every solution in the back section, never beside its puzzle."
+      },
+      {
+        "name": "delight",
+        "description": "The pack feels like a thing a family would fight over at the kitchen table."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "shelf completeness",
+    "solutions at the back",
+    "print-ready booklet"
+  ]
+}