@bendyline/gilde 0.1.37 → 0.1.39

package/data/craftbook-templates/br/broadcast-announcement/versions/1.0.2/craftbook.json

@@ -0,0 +1,119 @@
+{
+  "id": "broadcast-announcement",
+  "name": "Multi-Channel Broadcast Announcement",
+  "description": "Adapt a single announcement (launch, policy change, maintenance window, org news) into channel-tailored variants — email, Slack/Teams post, status-page note, and a short SMS/push — that carry one consistent core message at the right length and tone for each surface. Scopes the core message and the per-channel constraints FIRST (the single source of truth, the CTA, character/format limits and tone per channel, the must-include facts), then a copywriter writes every variant from that locked core so nothing contradicts across channels. Scoping the core before adapting is what keeps the facts, dates, and CTA identical everywhere while the voice fits each surface. Use for announcements, multi-channel comms, launch messaging, maintenance notices, and broadcasting one message everywhere.\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 message (planner) — lock the core message, CTA, and per-channel constraints → gated on artifact `{{workPath}}/scope.md` (markdown-notes)\n2. Write the variants (copywriter) — adapt the core into channel-tailored variants → gated on `announcement.md` (markdown-doc)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "multi-channel announcement",
+    "broadcast a message",
+    "launch announcement",
+    "maintenance notice",
+    "announce this everywhere"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Scope the message",
+      "description": "lock the core message, CTA, and per-channel constraints",
+      "prompt": "Define the announcement's single source of truth before writing variants. (1) Write the CORE message: the one thing being announced, the must-include facts (what, when/effective date, who's affected, why), and the single CTA. (2) Lock the channel list and each channel's constraints: email (subject + ~150 words, formal-friendly), Slack/Teams (~60 words, conversational, threaded details), status-page note (~40 words, neutral/factual), SMS/push (≤160 chars, urgent-but-calm, link only). (3) State the tone per channel and any compliance/do-not-say constraints. Write an acceptance-criteria checklist ('a variant exists for every channel', 'the core facts/dates/CTA are identical across all variants', 'each variant respects its length/format limit', 'each variant matches its channel tone', 'no variant contradicts another') plus the core message + per-channel constraint table to {{workPath}}/scope.md via write_task_note AND the produces path. No variants 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": "write"
+    },
+    {
+      "id": "write",
+      "name": "Write the variants",
+      "description": "adapt the core into channel-tailored variants",
+      "prompt": "Write announcement.md with one labeled section per channel, all derived from the locked core. (1) Email: a subject line plus a ~150-word body covering all must-include facts and the CTA, formal-friendly. (2) Slack/Teams: ~60 conversational words with the headline + CTA and a note that details follow. (3) Status-page: ~40 neutral, factual words stating what/when/impact. (4) SMS/push: ≤160 characters, the essential fact + CTA link, calm. Keep the facts, the effective date, and the CTA WORD-FOR-WORD consistent (where the medium allows) across every variant — only length and voice change. Stay within each length limit; count characters for SMS. Invent no facts beyond the core. On a loop-back fix only the flagged variants. Write the path and the per-channel lengths to notes via write_task_note.",
+      "suggestedRole": "copywriter",
+      "advanceWhen": {
+        "file": "announcement.md",
+        "minBytes": 1,
+        "sniff": "nonempty"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "announcement.md",
+            "bytes": 500
+          }
+        ],
+        "scripts": [
+          {
+            "name": "checkContains",
+            "scope": "standard",
+            "inputs": {
+              "file": "announcement.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 announcement.md against the locked criteria. (1) Confirm a variant exists for every channel in scope. (2) Cross-check the core facts, the effective date, and the CTA across all variants — any contradiction or drifted date/CTA is FAIL. (3) Check each variant against its length/format limit (count the SMS characters, ~word counts for the others). (4) Read each for channel-appropriate tone. (5) Confirm no fact was invented beyond the locked core. Write PASS/FAIL per criterion with the specific variants at fault.\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": "announcement.md"
+        }
+      ],
+      "next": "write"
+    },
+    {
+      "id": "finish",
+      "name": "Finish",
+      "description": "All acceptance criteria met. Stamp a short summary and report DONE.",
+      "prompt": "Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.",
+      "suggestedRole": "developer",
+      "terminal": true
+    }
+  ],
+  "version": "1.0.2",
+  "releasedAt": "2026-08-21T00:00:00Z",
+  "paramSchema": {
+    "type": "object",
+    "properties": {
+      "workPath": {
+        "type": "string",
+        "title": "Working folder",
+        "description": "Per-task working folder in the artifacts drawer. Defaults to this task's own folder so runs never collide; override with a stable name when you deliberately want runs to share files.",
+        "default": "{{task.dir}}"
+      }
+    }
+  },
+  "minGezelVersion": "1.26233"
+}

package/data/craftbook-templates/br/broadcast-announcement/versions/1.0.2/test.json

@@ -0,0 +1,106 @@
+{
+  "schemaVersion": 1,
+  "title": "Multi-channel announcement variants",
+  "objective": "Measure whether the broadcast-announcement craftbook locks a single source of truth before adapting channel-specific variants.",
+  "tags": [
+    "html-page"
+  ],
+  "prompt": "In the `Broadcast Announcement Eval` project, read `source/message-brief.md` and write `announcement.md` with Email, Slack/Teams, Status page, and SMS/Push variants. Keep the effective date and CTA consistent across all variants. Do not include a restrictions, prohibited-words, or checklist section in the deliverable; silently omit any forbidden terms.",
+  "setup": {
+    "projectName": "Broadcast Announcement Eval",
+    "about": "Self-contained eval project for broadcast-announcement. The core message is in workspace/source/message-brief.md.",
+    "missionObjectives": "Create channel-tailored announcement variants with identical facts, date, CTA, and no contradictions.",
+    "files": [
+      {
+        "path": "source/message-brief.md",
+        "content": "# Core message\n\nAnnouncement: HarborCRM is launching guided returns intake for all Pro accounts.\nEffective date: July 15, 2026 at 09:00 Pacific.\nAudience: support leads and account admins.\nImpact: customers will submit order number, reason code, and photo evidence before an agent picks up the case.\nCTA: review the rollout checklist at https://status.boreal.example/returns.\nDo not say: no-cost, instant, guaranteed, or beta.\n"
+      }
+    ],
+    "worker": {
+      "name": "Eline",
+      "role": "Copywriter"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "workspace/announcement.md contains four channel variants with consistent core facts, date, CTA, and forbidden claims excluded.",
+    "deliverables": [
+      {
+        "path": "announcement.md",
+        "kind": "markdown-doc",
+        "minBytes": 1800,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "announcement.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "announcement.md",
+            "pattern": "Email|Slack|Teams|Status|SMS|Push",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "announcement.md",
+            "pattern": "July 15, 2026|09:00 Pacific",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "announcement.md",
+            "pattern": "https://status\\.boreal\\.example/returns",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "announcement.md",
+            "pattern": "guided returns intake|order number|reason code|photo evidence",
+            "flags": "i"
+          },
+          {
+            "kind": "notContains",
+            "file": "announcement.md",
+            "pattern": "no-cost|instant|guaranteed|beta",
+            "flags": "i",
+            "label": "forbidden launch claims must not appear anywhere, including restrictions, checklists, or meta notes"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "announcement.md",
+      "kind": "markdown"
+    },
+    "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": [
+    "single source of truth",
+    "channel adaptation",
+    "consistency checks"
+  ]
+}

package/data/craftbook-templates/br/browser-qa-audit/versions/2.0.4/craftbook.json

@@ -0,0 +1,429 @@
+{
+  "id": "browser-qa-audit",
+  "name": "Browser QA Audit",
+  "description": "Exercise a running app in a real browser, triage what breaks, and produce a report-only QA findings list with a health score — no code changes.",
+  "basedOn": {
+    "name": "gstack",
+    "url": "https://github.com/garrytan/gstack"
+  },
+  "plan": "Perform black-box, report-only browser QA. Do not read implementation code and do not fix defects. Test the supplied running target as a real user across core journeys, navigation, forms, validation, state changes, console health, responsive layouts, and accessibility basics. Capture evidence at the time of observation, retry suspected defects once, redact credentials, and distinguish tested, blocked, and out-of-scope surfaces. A health score is only valid when its evidence and arithmetic are inspectable.",
+  "entryStepId": "plan-coverage",
+  "triggers": [
+    "qa report only",
+    "just report bugs",
+    "test but dont fix"
+  ],
+  "command": "browser-qa-audit",
+  "toolsets": [
+    {
+      "toolsetId": "@playwright/mcp",
+      "optional": false,
+      "reason": "exercise the target through a real browser and capture reproducible evidence"
+    }
+  ],
+  "steps": [
+    {
+      "id": "plan-coverage",
+      "name": "Plan observable coverage",
+      "description": "Confirm the target and turn user intent into bounded journeys, states, and evidence requirements.",
+      "prompt": "Confirm the target URL, requested scope, authentication constraints, viewport requirements, destructive-action limits, and expected behavior from the task context. Ask only for missing information that blocks browser use, and never store credentials in output. Open the target with the configured browser toolset to verify reachability. Map the visible navigation and prioritize core journeys. Define happy, empty, invalid, error, loading, permission, and responsive states to exercise, plus a clear tested versus excluded boundary.\n\nObservable handoff: write the completed result to `qa/test-plan.md` in the workspace with `write_file`. Do not merely describe what the file would contain. Re-read it with `read_file` before finishing this phase and repair any incomplete sections.",
+      "suggestedRole": "QA analyst",
+      "advanceWhen": {
+        "file": "qa/test-plan.md",
+        "minBytes": 700,
+        "sniff": "nonempty",
+        "requireChange": true,
+        "goto": "exercise"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "qa/test-plan.md",
+            "bytes": 700
+          },
+          {
+            "kind": "sniff",
+            "file": "qa/test-plan.md",
+            "sniff": "nonempty"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/test-plan.md",
+            "pattern": "^##\\s+Target",
+            "flags": "im",
+            "label": "Target section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/test-plan.md",
+            "pattern": "^##\\s+Core journeys",
+            "flags": "im",
+            "label": "Core journeys section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/test-plan.md",
+            "pattern": "^##\\s+State matrix",
+            "flags": "im",
+            "label": "State matrix section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/test-plan.md",
+            "pattern": "^##\\s+Viewports",
+            "flags": "im",
+            "label": "Viewports section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/test-plan.md",
+            "pattern": "^##\\s+Scope boundaries",
+            "flags": "im",
+            "label": "Scope boundaries section"
+          }
+        ],
+        "onReject": "plan-coverage",
+        "maxAttempts": 3
+      },
+      "next": "exercise"
+    },
+    {
+      "id": "exercise",
+      "name": "Exercise the application",
+      "description": "Run the planned journeys in a real browser and capture reproducible, redacted evidence.",
+      "prompt": "Execute `qa/test-plan.md` with the configured browser tools, using `run_playwright_script` for repeatable multi-step journeys when useful. At every journey, record URL, viewport, input state, actions, visible result, console errors, and screenshot path. Test keyboard focus and basic accessible naming, realistic invalid input, browser back and forward behavior, and at least one narrow viewport. Retry each suspected issue once. Save screenshots beneath `qa/screenshots/` and a machine-readable issue summary to `qa/baseline.json`. Redact secrets and never submit destructive or externally visible actions without explicit permission.\n\nObservable handoff: write the completed result to `qa/browser-evidence.md` in the workspace with `write_file`. Do not merely describe what the file would contain. Re-read it with `read_file` before finishing this phase and repair any incomplete sections.",
+      "suggestedRole": "browser QA engineer",
+      "consumes": [
+        {
+          "file": "qa/test-plan.md"
+        }
+      ],
+      "advanceWhen": {
+        "file": "qa/browser-evidence.md",
+        "minBytes": 1200,
+        "sniff": "nonempty",
+        "requireChange": true,
+        "goto": "report-findings"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "qa/browser-evidence.md",
+            "bytes": 1200
+          },
+          {
+            "kind": "sniff",
+            "file": "qa/browser-evidence.md",
+            "sniff": "nonempty"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-evidence.md",
+            "pattern": "^##\\s+Run metadata",
+            "flags": "im",
+            "label": "Run metadata section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-evidence.md",
+            "pattern": "^##\\s+Journey results",
+            "flags": "im",
+            "label": "Journey results section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-evidence.md",
+            "pattern": "^##\\s+Console",
+            "flags": "im",
+            "label": "Console section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-evidence.md",
+            "pattern": "^##\\s+Responsive and accessibility",
+            "flags": "im",
+            "label": "responsive and accessibility section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-evidence.md",
+            "pattern": "^##\\s+Evidence index",
+            "flags": "im",
+            "label": "Evidence index section"
+          },
+          {
+            "kind": "sniff",
+            "file": "qa/baseline.json",
+            "sniff": "json-valid"
+          }
+        ],
+        "onReject": "exercise",
+        "maxAttempts": 3
+      },
+      "next": "report-findings"
+    },
+    {
+      "id": "report-findings",
+      "name": "Write the QA report",
+      "description": "Triage observed defects, compute the health score, and provide owners with exact reproduction evidence.",
+      "prompt": "Write the report only from `qa/test-plan.md`, `qa/browser-evidence.md`, screenshots, and baseline data. For every issue include severity, confidence, category, affected journey, exact reproduction steps, expected versus actual behavior, screenshot or console evidence, frequency, and user impact. Keep blocked checks separate from passes. Compute category scores and the weighted health score with visible arithmetic. End with the top three fixes by user impact, but do not edit code or prescribe speculative implementation details.\n\nObservable handoff: write the completed result to `qa/browser-qa-report.md` in the workspace with `write_file`. Do not merely describe what the file would contain. Re-read it with `read_file` before finishing this phase and repair any incomplete sections.",
+      "suggestedRole": "QA reporter",
+      "consumes": [
+        {
+          "file": "qa/test-plan.md"
+        },
+        {
+          "file": "qa/browser-evidence.md"
+        }
+      ],
+      "advanceWhen": {
+        "file": "qa/browser-qa-report.md",
+        "minBytes": 1500,
+        "sniff": "nonempty",
+        "requireChange": true,
+        "goto": "evaluate"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "qa/browser-qa-report.md",
+            "bytes": 1500
+          },
+          {
+            "kind": "sniff",
+            "file": "qa/browser-qa-report.md",
+            "sniff": "nonempty"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Executive summary",
+            "flags": "im",
+            "label": "Executive summary section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Coverage",
+            "flags": "im",
+            "label": "Coverage section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Findings",
+            "flags": "im",
+            "label": "Findings section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Console health",
+            "flags": "im",
+            "label": "Console health section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Health score",
+            "flags": "im",
+            "label": "Health score section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Top three fixes",
+            "flags": "im",
+            "label": "Top three fixes section"
+          }
+        ],
+        "onReject": "report-findings",
+        "maxAttempts": 3
+      },
+      "next": "evaluate"
+    },
+    {
+      "id": "evaluate",
+      "name": "Evaluate the deliverable",
+      "description": "Independently grade the observable deliverable and route it to finish, repair, or user escalation.",
+      "prompt": "Review `qa/browser-qa-report.md`, `qa/test-plan.md`, `qa/browser-evidence.md`, `qa/baseline.json` against every criterion below. Inspect the underlying evidence files named by the workflow; do not grade from the author's summary alone.\n\n1. Every claimed pass, fail, and blocked check traces to an actual browser journey in the evidence log.\n2. Each defect was reproduced twice or its confidence was reduced, and includes expected versus actual behavior plus screenshot or console evidence.\n3. Core journeys cover meaningful state changes, invalid input, navigation, console health, responsive layout, and accessibility basics in proportion to scope.\n4. Credentials and sensitive data are redacted, and no destructive or externally visible action was taken without permission.\n5. Severity reflects user impact and reach, while untested surfaces are never counted as passing.\n6. The category and overall health-score arithmetic is reproducible from the report.\n\nOpen `qa/browser-qa-report.md`, `qa/test-plan.md`, `qa/browser-evidence.md`, `qa/baseline.json` with `read_file`. Write the evidence-backed review to `{{workPath}}/reviews/browser-qa-audit-review.md` in the artifacts drawer with `write_artifact`. Give each criterion a PASS or FAIL with a concrete path, excerpt, measurement, or observed behavior. End with exactly `Verdict: PASS` or `Verdict: REVISE`. Then use `advance_task_step` for the active task: PASS routes to `finish`; REVISE routes to `repair` for review rounds 1 through 2, and the 3th REVISE routes to `needs-user`. Never route to finish while a criterion is unmet.",
+      "suggestedRole": "QA reviewer",
+      "consumes": [
+        {
+          "file": "qa/browser-qa-report.md"
+        },
+        {
+          "file": "qa/test-plan.md"
+        },
+        {
+          "file": "qa/browser-evidence.md"
+        },
+        {
+          "file": "qa/baseline.json"
+        }
+      ],
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/reviews/browser-qa-audit-review.md",
+            "bytes": 400,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/reviews/browser-qa-audit-review.md",
+            "pattern": "Verdict:\\s*(?:PASS|REVISE)",
+            "flags": "i",
+            "label": "explicit PASS or REVISE verdict",
+            "artifact": true
+          }
+        ],
+        "onReject": "evaluate",
+        "maxAttempts": 3
+      },
+      "next": "repair"
+    },
+    {
+      "id": "repair",
+      "name": "Repair the deliverable",
+      "description": "Fix only the concrete gaps from the latest independent review.",
+      "prompt": "Read `{{workPath}}/reviews/browser-qa-audit-review.md` with `read_artifact` and repair every failed criterion in `qa/browser-qa-report.md`, `qa/test-plan.md`, `qa/browser-evidence.md`, `qa/baseline.json`. Make changes on each file's declared surface (`write_artifact` for artifact inputs, `write_file` for workspace inputs), not in task notes or a reply. Preserve evidence that already passed. Re-run or re-check anything the reviewer found unproven. Ensure `qa/browser-qa-report.md` is genuinely updated this turn so the repair is observable, then hand it back for independent evaluation.",
+      "suggestedRole": "QA reporter",
+      "consumes": [
+        {
+          "file": "{{workPath}}/reviews/browser-qa-audit-review.md",
+          "artifact": true
+        },
+        {
+          "file": "qa/browser-qa-report.md"
+        },
+        {
+          "file": "qa/test-plan.md"
+        },
+        {
+          "file": "qa/browser-evidence.md"
+        },
+        {
+          "file": "qa/baseline.json"
+        }
+      ],
+      "advanceWhen": {
+        "file": "qa/browser-qa-report.md",
+        "minBytes": 1500,
+        "sniff": "nonempty",
+        "requireChange": true,
+        "goto": "evaluate"
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "qa/browser-qa-report.md",
+            "bytes": 1500
+          },
+          {
+            "kind": "sniff",
+            "file": "qa/browser-qa-report.md",
+            "sniff": "nonempty"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Executive summary",
+            "flags": "im",
+            "label": "Executive summary section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Coverage",
+            "flags": "im",
+            "label": "Coverage section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Findings",
+            "flags": "im",
+            "label": "Findings section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Console health",
+            "flags": "im",
+            "label": "Console health section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Health score",
+            "flags": "im",
+            "label": "Health score section"
+          },
+          {
+            "kind": "contains",
+            "file": "qa/browser-qa-report.md",
+            "pattern": "^##\\s+Top three fixes",
+            "flags": "im",
+            "label": "Top three fixes section"
+          }
+        ],
+        "onReject": "repair",
+        "maxAttempts": 3
+      },
+      "next": "evaluate"
+    },
+    {
+      "id": "finish",
+      "name": "Finish",
+      "description": "All deterministic and reviewer criteria passed.",
+      "prompt": "The independent review passed. Read `{{workPath}}/reviews/browser-qa-audit-review.md` with `read_artifact`, then use `write_task_note` to record a concise DONE summary with the final deliverable paths (`qa/browser-qa-report.md`, `qa/test-plan.md`, `qa/browser-evidence.md`, `qa/baseline.json`) and the evidence that each acceptance criterion passed. Report DONE without starting new work.",
+      "suggestedRole": "project lead",
+      "consumes": [
+        {
+          "file": "{{workPath}}/reviews/browser-qa-audit-review.md",
+          "artifact": true
+        }
+      ],
+      "terminal": true
+    },
+    {
+      "id": "needs-user",
+      "name": "Escalate unresolved concerns",
+      "description": "The bounded repair loop ended without a defensible pass.",
+      "prompt": "The deliverable did not pass after 3 review rounds. Do not claim success. Read `{{workPath}}/reviews/browser-qa-audit-review.md` with `read_artifact`, then use `write_task_note` to record DONE_WITH_CONCERNS: the unmet criteria, what was attempted, the affected paths, and the smallest user decision or missing input needed to continue.",
+      "suggestedRole": "project lead",
+      "consumes": [
+        {
+          "file": "{{workPath}}/reviews/browser-qa-audit-review.md",
+          "artifact": true
+        }
+      ],
+      "terminal": true
+    }
+  ],
+  "version": "2.0.4",
+  "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"
+}