npm - @bendyline/gilde - Versions diffs - 0.1.26 → 0.1.28 - Mend

@bendyline/gilde 0.1.26 → 0.1.28

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (571) hide show

package/data/craftbook-templates/de/design-system-consultation/versions/2.0.3/test.json ADDED Viewed

@@ -0,0 +1,201 @@
+{
+  "schemaVersion": 1,
+  "title": "Design System Consultation — Harbor Ops dashboard",
+  "objective": "Exercise the Design System Consultation craftbook on a seeded operations dashboard and require a grounded, accessible DESIGN.md plus a complete HTML state preview.",
+  "tags": [
+    "workflow",
+    "design-system",
+    "html-page",
+    "accessibility"
+  ],
+  "prompt": "Use the Design System Consultation craftbook for the Harbor Ops dashboard. This is an unattended evaluation: the product and taste decisions in `source/design-brief.md` are authoritative, so proceed without asking questions and label any remaining assumptions. Audit the supplied interface, preserve its useful density, then produce the workflow artifacts, `DESIGN.md`, and `previews/design-system.html`. The preview must be self-contained and show navigation, typography, controls, data cards, table content, empty/loading/error/success states, keyboard focus, disabled controls, and responsive behavior. Use no external assets or live research.",
+  "setup": {
+    "projectName": "Harbor Ops design system",
+    "about": "Harbor Ops is a desktop-first operations dashboard for coordinators monitoring time-sensitive berth assignments. The fixture is complete and must be handled without live research.",
+    "missionObjectives": "Run the Design System Consultation craftbook as a real task. Ground every design decision in the seeded brief and current interface, then deliver DESIGN.md and the self-contained preview.",
+    "files": [
+      {
+        "path": "README.md",
+        "content": "# Harbor Ops\n\nHarbor Ops helps port coordinators scan berth conflicts, assign an owner, and clear exceptions during a shift. It is dense desktop software used in bright control rooms; the core job is spotting the one item that needs action now.\n"
+      },
+      {
+        "path": "source/design-brief.md",
+        "content": "# Locked design brief\n\nAudience: experienced port coordinators working 8-hour shifts on desktop, with occasional tablet use.\n\nMemorable intent: **calm command under pressure**. The interface should feel serious and precise, never playful or luxurious.\n\nPreserve: dense scan-friendly tables, the navy header, status labels, and explicit timestamps.\n\nChange: inconsistent spacing, low-contrast muted text, color-only severity, invisible keyboard focus, and one-off button styles.\n\nBrand constraints: deep navy `#10233f`; signal amber `#d97706`; off-white canvas. Do not use pure black.\n\nAccessibility: WCAG AA text contrast, visible `:focus-visible`, shape/text in addition to status color, 44px primary touch targets on tablet, reduced-motion support.\n\nProduct states that must be designed: normal, loading, empty, permission error, delayed-data warning, success acknowledgement, selected row, disabled action.\n\nTaste: compact Swiss/industrial typography, restrained motion, square-to-soft corners (4-8px), clear dividers. Avoid gradients, glass effects, giant hero type, and decorative illustrations.\n"
+      },
+      {
+        "path": "src/dashboard.html",
+        "content": "<!doctype html><html lang=\"en\"><head><meta charset=\"utf-8\"><title>Harbor Ops</title><link rel=\"stylesheet\" href=\"styles.css\"></head><body><header>Harbor Ops <small>Shift A · 07:42 UTC</small></header><main><aside><button>Overview</button><button>Berths</button><button>Exceptions</button></aside><section><h1>Berth exceptions</h1><button class=\"primary\">Assign owner</button><table><tr><th>Vessel</th><th>Berth</th><th>Status</th><th>Updated</th></tr><tr><td>MV Juniper</td><td>B-14</td><td><span class=\"status\">Delayed</span></td><td>07:38 UTC</td></tr></table></section></main></body></html>\n"
+      },
+      {
+        "path": "src/styles.css",
+        "content": ":root{--navy:#10233f;--amber:#d97706}body{margin:0;font:14px Arial;background:#f7f5ef;color:#27364a}header{background:var(--navy);color:white;padding:13px 18px}header small{color:#7e8aa0}main{display:flex}aside{width:190px;padding:12px}section{flex:1;padding:19px}button{border:0;padding:7px;margin:3px}.primary{background:var(--amber);color:white}.status{color:#d97706}table{width:100%;border-collapse:collapse}td,th{text-align:left;border-bottom:1px solid #ddd;padding:9px}button:focus{outline:none}\n"
+      }
+    ]
+  },
+  "mocks": [],
+  "success": {
+    "summary": "The real craftbook produces a grounded design audit and brief, an accessible token specification, a valid multi-state preview, and terminal review evidence.",
+    "deliverables": [
+      {
+        "path": "design/current-state-audit.md",
+        "kind": "markdown-report",
+        "minBytes": 750,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "design/current-state-audit.md",
+            "pattern": "^#{1,3}\\s+Product and users\\b[\\s\\S]*^#{1,3}\\s+Existing language\\b[\\s\\S]*^#{1,3}\\s+Accessibility\\b",
+            "flags": "im",
+            "label": "grounded current-state audit"
+          },
+          {
+            "kind": "citationsResolve",
+            "file": "design/current-state-audit.md",
+            "minCitations": 2
+          }
+        ]
+      },
+      {
+        "path": "DESIGN.md",
+        "kind": "markdown-doc",
+        "minBytes": 1800,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "DESIGN.md",
+            "pattern": "^#{1,3}\\s+Design intent\\b[\\s\\S]*^#{1,3}\\s+Tokens\\b[\\s\\S]*^#{1,3}\\s+Typography\\b[\\s\\S]*^#{1,3}\\s+Color\\b[\\s\\S]*^#{1,3}\\s+Components and states\\b[\\s\\S]*^#{1,3}\\s+Accessibility\\b[\\s\\S]*^#{1,3}\\s+Adoption\\b",
+            "flags": "im",
+            "label": "complete design-system specification"
+          },
+          {
+            "kind": "valueGrounding",
+            "file": "DESIGN.md",
+            "facts": [
+              {
+                "id": "intent",
+                "label": "memorable intent",
+                "required": [
+                  "calm command under pressure"
+                ]
+              },
+              {
+                "id": "brand-navy",
+                "label": "locked navy",
+                "required": [
+                  "#10233f"
+                ]
+              },
+              {
+                "id": "brand-amber",
+                "label": "locked signal amber",
+                "required": [
+                  "#d97706"
+                ]
+              },
+              {
+                "id": "keyboard-focus",
+                "label": "visible keyboard focus",
+                "required": [
+                  "focus-visible"
+                ]
+              },
+              {
+                "id": "reduced-motion",
+                "label": "reduced-motion behavior",
+                "required": [
+                  "reduced-motion|prefers-reduced-motion"
+                ]
+              },
+              {
+                "id": "touch-target",
+                "label": "tablet touch target",
+                "required": [
+                  "44\\s*(?:px|pixel)"
+                ]
+              }
+            ]
+          }
+        ]
+      },
+      {
+        "path": "previews/design-system.html",
+        "kind": "html-page",
+        "minBytes": 3000,
+        "checks": [
+          {
+            "kind": "htmlLint",
+            "file": "previews/design-system.html"
+          },
+          {
+            "kind": "cssMinBytes",
+            "bytes": 900,
+            "file": "previews/design-system.html"
+          },
+          {
+            "kind": "contains",
+            "file": "previews/design-system.html",
+            "pattern": "loading[\\s\\S]*empty[\\s\\S]*(error|permission)[\\s\\S]*(success|acknowledge)[\\s\\S]*disabled",
+            "flags": "i",
+            "label": "representative application states"
+          },
+          {
+            "kind": "contains",
+            "file": "previews/design-system.html",
+            "pattern": "focus-visible|prefers-reduced-motion",
+            "flags": "i",
+            "label": "keyboard focus and reduced motion"
+          }
+        ]
+      }
+    ],
+    "taskNotes": {
+      "minBytes": 180,
+      "checks": [
+        {
+          "kind": "contains",
+          "file": "task-notes.md",
+          "pattern": "\\bDONE\\b[\\s\\S]*DESIGN\\.md[\\s\\S]*previews/design-system\\.html",
+          "flags": "i",
+          "label": "terminal note records both design deliverables"
+        }
+      ],
+      "requireCraftbookTask": true
+    },
+    "taskGraph": {
+      "requireCraftbookTask": true,
+      "requireTerminalStep": true
+    },
+    "unchangedFixtures": [
+      "README.md",
+      "source/design-brief.md",
+      "src/dashboard.html",
+      "src/styles.css"
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "DESIGN.md",
+      "kind": "markdown"
+    },
+    "axes": [
+      {
+        "name": "Product fit",
+        "description": "The system reinforces calm command under pressure and the actual coordinator workflows rather than generic visual taste."
+      },
+      {
+        "name": "System coherence",
+        "description": "Tokens, typography, color, density, components, states, and motion form a reusable system with explicit rationale."
+      },
+      {
+        "name": "Accessibility and preview",
+        "description": "The specification and preview cover contrast, focus, non-color status, touch size, reduced motion, responsiveness, and real application states."
+      }
+    ],
+    "contextNote": "Judge against the seeded brief and current dashboard; decorative novelty that violates the locked taste is a defect."
+  },
+  "qualityFocus": [
+    "Preserve useful density while correcting current accessibility failures",
+    "Use exact reusable tokens with reasons",
+    "Preview representative responsive and interaction states without external assets"
+  ]
+}

package/data/craftbook-templates/di/diagram-from-text/versions/1.0.1/craftbook.json ADDED Viewed

@@ -0,0 +1,109 @@
+{
+  "id": "diagram-from-text",
+  "name": "Diagram from Text",
+  "description": "Render a clear technical diagram — flowchart, architecture, sequence, ER, or org chart — from a plain-language description, as code-defined (Mermaid/Graphviz/SVG) so it is editable and version-controllable rather than a flat picture. Spec FIRST the diagram type, the exact nodes and edges, and the layout direction, because diagrams fail when relationships are wrong or the graph is an unreadable tangle; then build the diagram so it renders, with every described entity and relationship present and labeled. Covers flowcharts, architecture diagrams, sequence diagrams, ER diagrams, org charts, mind maps, and diagram-as-code.\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. Spec the diagram (planner) — lock type, nodes, edges, labels, and layout → gated on artifact `notes/spec.md` (markdown-notes)\n2. Build the diagram (developer) — author the renderable diagram to the spec → gated on `index.html` (html-page)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "spec",
+  "triggers": [
+    "make a diagram from this description",
+    "draw a flowchart",
+    "architecture diagram",
+    "sequence diagram",
+    "turn text into a diagram"
+  ],
+  "steps": [
+    {
+      "id": "spec",
+      "name": "Spec the diagram",
+      "description": "lock type, nodes, edges, labels, and layout",
+      "prompt": "1) Parse the description and choose the right diagram TYPE (flowchart, sequence, ER, class, state, org chart, mind map). 2) Enumerate the exact NODES (entities/steps/actors) with their labels and the exact EDGES (relationships/transitions/messages) with their labels and direction — list them explicitly so nothing is dropped or invented. 3) Decide layout direction (top-down vs left-right) and any grouping/subgraphs. 4) Pick the rendering format (Mermaid is the default for editability; Graphviz/SVG if requested). 5) Write an acceptance-criteria checklist: 'every entity in the description is a node', 'every stated relationship is an edge with correct direction', 'labels match the description', 'the diagram renders without syntax errors', 'layout is readable (no needless crossings)'. 6) Record the node/edge list + checklist via write_task_note AND to notes/spec.md.\n\nThe deliverable `notes/spec.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": "notes/spec.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "notes/spec.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "notes/spec.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "spec",
+        "maxAttempts": 3
+      },
+      "next": "build"
+    },
+    {
+      "id": "build",
+      "name": "Build the diagram",
+      "description": "author the renderable diagram to the spec",
+      "prompt": "1) Read spec.md's node/edge list. 2) Author the diagram in the chosen format (e.g. a Mermaid graph in an index.html that loads mermaid.js and renders it, so it actually displays). 3) Add EVERY node and EVERY edge from the spec with the correct labels and direction — do not add entities that weren't described and do not omit any. 4) Set the layout direction and use subgraphs/grouping for clarity; choose an arrangement that minimizes edge crossings. 5) Verify the syntax parses (the diagram renders, no error box). 6) On a loop-back, fix only the named gaps (missing edge, wrong arrow direction, syntax error). 7) write_task_note the file path and node/edge counts. The deliverable is index.html that renders the diagram.",
+      "suggestedRole": "developer",
+      "advanceWhen": {
+        "file": "index.html",
+        "minBytes": 1,
+        "sniff": "html-complete"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "index.html",
+            "bytes": 800
+          },
+          {
+            "kind": "htmlLint",
+            "file": "index.html"
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkHtmlComplete",
+            "scope": "standard",
+            "inputs": {
+              "file": "index.html"
+            }
+          }
+        ],
+        "onReject": "build",
+        "maxAttempts": 4
+      },
+      "next": "evaluate"
+    },
+    {
+      "id": "evaluate",
+      "name": "Evaluate",
+      "description": "Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.",
+      "prompt": "1) Open index.html and confirm the diagram actually RENDERS with no syntax/error box (PASS/FAIL). 2) Cross-check against spec.md: every described entity appears as a node and every stated relationship appears as an edge with the correct DIRECTION and label (PASS/FAIL — this is the core check). 3) Confirm no invented nodes/edges. 4) Judge readability — is the layout legible or a tangle? 5) Write PASS/FAIL per criterion. Missing/wrong relationships or a non-rendering diagram → loop back to build.\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: \"build\" })` to loop back. The builder fixes exactly those gaps.\n\nNever route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in.",
+      "suggestedRole": "reviewer",
+      "consumes": [
+        {
+          "file": "index.html"
+        }
+      ],
+      "next": "build"
+    },
+    {
+      "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": "1.0.1",
+  "releasedAt": "2026-08-14T00:00:00Z"
+}

package/data/craftbook-templates/di/diagram-from-text/versions/1.0.1/test.json ADDED Viewed

@@ -0,0 +1,97 @@
+{
+  "schemaVersion": 1,
+  "title": "Diagram from Text smoke eval",
+  "objective": "Self-contained smoke eval for the Diagram from Text craftbook using the html-page generic harness.",
+  "tags": [
+    "html-page"
+  ],
+  "prompt": "Can you biuld us a little page for this? The content notes are in source/page-content.md. Save it as index.html — one file, nothing fancy needed on our end.",
+  "setup": {
+    "projectName": "Diagram from Text Eval",
+    "about": "Self-contained eval project for diagram-from-text. Seeded inputs are under workspace/source or workspace/fixtures; final deliverable is workspace/index.html.",
+    "missionObjectives": "Use the Diagram from Text craftbook/template, read the seeded local fixtures, and write index.html without network calls, real credentials, or live services.",
+    "files": [
+      {
+        "path": "source/brief.md",
+        "content": "# Diagram from Text Eval Brief\n\nClient: Boreal Desk, a home-office accessories company.\nAudience: operations leads who need an artifact they can use this week.\n\nFixed source facts for grounding:\n- The returns desk pilot covered 18 SKUs.\n- Median first response improved from 18 hours to 6 hours.\n- Preventable refund leakage fell from 14.2% to 8.9%.\n- The top unresolved complaint is status silence after photo submission.\n- Required next actions are automated status emails, barcode-exception training, and a weekly Finance exception export.\n\nUse these facts when the task asks for prose, analysis, copy, UI content, or test data. Do not use live web services, real credentials, or current outside data.\n\nCraftbook under test: diagram-from-text - Diagram from Text.\n"
+      },
+      {
+        "path": "source/page-content.md",
+        "content": "# Boreal Desk Support Board\n\nBuild content around a support operations board. Required labels: Open issues, Status silence, Response time, Refund leakage, Next action. Required metric text: 18 hours to 6 hours and 14.2% to 8.9%.\n"
+      }
+    ],
+    "worker": {
+      "name": "Jules",
+      "role": "Developer"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "index.html is a source-grounded self-contained HTML page/tool.",
+    "deliverables": [
+      {
+        "path": "index.html",
+        "kind": "html-page",
+        "minBytes": 3000,
+        "checks": [
+          {
+            "kind": "cssMinBytes",
+            "bytes": 650,
+            "file": "index.html"
+          },
+          {
+            "kind": "jsParses",
+            "file": "index.html"
+          },
+          {
+            "kind": "contains",
+            "file": "index.html",
+            "pattern": "Boreal|Support Board|returns",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "index.html",
+            "pattern": "18\\s*hours|6\\s*hours|14\\.2%|8\\.9%",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "index.html",
+            "pattern": "button|input|filter|tab|card|section|nav",
+            "flags": "i"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "index.html",
+      "kind": "html"
+    },
+    "axes": [
+      {
+        "name": "structure",
+        "description": "Layout and hierarchy fit the content; sections are scannable and coherent."
+      },
+      {
+        "name": "grounding",
+        "description": "Seeded source content is used faithfully — no invented facts or dropped requirements."
+      },
+      {
+        "name": "interactivity",
+        "description": "Controls and states the page claims to offer actually work."
+      },
+      {
+        "name": "polish",
+        "description": "Styling is intentional and responsive rather than unstyled or broken."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "self-contained UI",
+    "source grounding",
+    "responsive structure"
+  ]
+}

package/data/craftbook-templates/di/digest-from-feeds/versions/1.0.1/craftbook.json ADDED Viewed

@@ -0,0 +1,146 @@
+{
+  "id": "digest-from-feeds",
+  "name": "RSS / Feed Digest",
+  "description": "Build a recurring digest from RSS/Atom feeds or news sources: fetch the latest items in the window, dedupe and cluster by topic, then write a scannable digest with one-line takes and source links. Scopes the feed list and digest format FIRST (which feeds, the time window, topic sections, item cap, ranking signal), then a developer fetches and normalizes the feed items into a deduped item list, then a copywriter writes the curated digest from those items only. Scoping sources and the window before fetching is what keeps the digest fresh, deduped, and within length instead of an unranked firehose. Use for RSS digests, news roundups, feed summaries, content curation, and recurring reading digests.\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. Source scope (planner) — lock the feeds, window, sections, and item cap → gated on artifact `notes/source-scope.md` (markdown-notes)\n2. Gather feed items (developer) — fetch, normalize, dedupe items into an item list → gated on artifact `notes/gather.md` (markdown-notes)\n3. Write the digest (copywriter) — compose the curated digest from gathered items → gated on `digest.md` (markdown-doc)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "source-scope",
+  "triggers": [
+    "rss digest",
+    "news roundup",
+    "feed summary",
+    "summarize these feeds",
+    "curated reading digest"
+  ],
+  "steps": [
+    {
+      "id": "source-scope",
+      "name": "Source scope",
+      "description": "lock the feeds, window, sections, and item cap",
+      "prompt": "Define the digest contract before fetching. (1) Lock the feed/source list (the RSS/Atom URLs or named sources). (2) Set the time window (only items published within it) and the topic sections the digest groups into. (3) Set a per-section item cap and an overall cap, plus a ranking signal for which items make the cut (recency, source weight, or keyword relevance to a stated interest). (4) Define the per-item format: a tight one-line take + a source link + publish date. (5) Define the dedupe rule (same story across feeds collapses to one, by URL or title similarity). Write an acceptance-criteria checklist ('only items within the window', 'duplicates across feeds collapsed', 'each item has a working source link + date', 'per-section and overall caps respected', 'every item traces to a fetched feed item — nothing invented', 'scannable in under two minutes') plus the source list + format to notes/source-scope.md via write_task_note AND the produces path.\n\nThe deliverable `notes/source-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": "notes/source-scope.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "notes/source-scope.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "notes/source-scope.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "source-scope",
+        "maxAttempts": 3
+      },
+      "next": "gather"
+    },
+    {
+      "id": "gather",
+      "name": "Gather feed items",
+      "description": "fetch, normalize, dedupe items into an item list",
+      "prompt": "Fetch and normalize the feeds into notes/gather.md. (1) Retrieve each feed (via the fetch/web tool); parse out item title, link, publish date, and a short excerpt. (2) Drop items outside the locked window. (3) Apply the dedupe rule — collapse the same story appearing in multiple feeds to a single entry, keeping the best source link. (4) Tag each surviving item with its target section and a ranking score per the locked signal. (5) Sort within each section by the ranking signal and keep within the caps; record the dropped count. Capture this as a structured, attributed list — title, link, date, section, score — NOT prose, and verify every link is a real URL from the feed. If a feed is unreachable, note it explicitly. Write the item list via write_task_note and end with per-section counts.\n\nThe deliverable `notes/gather.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": "developer",
+      "advanceWhen": {
+        "file": "notes/gather.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "notes/gather.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "notes/gather.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "gather",
+        "maxAttempts": 3
+      },
+      "next": "write"
+    },
+    {
+      "id": "write",
+      "name": "Write the digest",
+      "description": "compose the curated digest from gathered items",
+      "prompt": "Write digest.md from notes/gather.md only. (1) Open with a one-line 'what mattered' summary of the period. (2) One section per topic in the locked order; under each, the capped items as a one-line take + linked title + date. Each take is a genuine, specific summary of the item — not a restated headline, not invented detail. (3) Respect the per-section and overall caps; if a section has no in-window items, omit it. (4) Keep the whole thing skimmable in under two minutes. Use only items present in the gather file with their real links — invent nothing. On a loop-back fix only the flagged sections. Write the path and item count to notes via write_task_note.\n\nBefore working, open the artifact input `notes/gather.md` with `read_artifact`; do not look for it in the workspace.",
+      "suggestedRole": "copywriter",
+      "consumes": [
+        {
+          "file": "notes/gather.md",
+          "artifact": true
+        }
+      ],
+      "advanceWhen": {
+        "file": "digest.md",
+        "minBytes": 1,
+        "sniff": "nonempty"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "digest.md",
+            "bytes": 500
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkContains",
+            "scope": "standard",
+            "inputs": {
+              "file": "digest.md",
+              "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+              "flags": "i"
+            }
+          }
+        ],
+        "onReject": "write",
+        "maxAttempts": 4
+      },
+      "next": "evaluate"
+    },
+    {
+      "id": "evaluate",
+      "name": "Evaluate",
+      "description": "Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.",
+      "prompt": "QA digest.md against the locked criteria. (1) Confirm every item appears in notes/gather.md with a real source link and an in-window date; FAIL on any invented item or out-of-window entry. (2) Confirm duplicates across feeds were collapsed (no two entries for the same story). (3) Confirm per-section and overall caps are respected. (4) Read 3 one-line takes — are they real summaries of the linked item, not just the headline echoed, and accurate? (5) Confirm it is genuinely scannable. Write PASS/FAIL per criterion with specifics.\n\nThen route — this is the whole point of the loop:\n\n- **Every criterion PASSES →** call `advance_task_step({ ref, stepId: \"evaluate\", next: \"finish\" })`.\n- **Any criterion FAILS →** write the specific gaps to notes, then call `advance_task_step({ ref, stepId: \"evaluate\", next: \"write\" })` to loop back. The builder fixes exactly those gaps.\n\nNever route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in.",
+      "suggestedRole": "reviewer",
+      "consumes": [
+        {
+          "file": "digest.md"
+        }
+      ],
+      "next": "write"
+    },
+    {
+      "id": "finish",
+      "name": "Finish",
+      "description": "All acceptance criteria met. Stamp a short summary and report DONE.",
+      "prompt": "Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.",
+      "suggestedRole": "developer",
+      "terminal": true
+    }
+  ],
+  "version": "1.0.1",
+  "releasedAt": "2026-08-14T00:00:00Z"
+}