@bendyline/gilde 0.1.38 → 0.1.40

package/data/craftbook-templates/co/cohort-analysis/versions/1.0.2/test.json

@@ -0,0 +1,229 @@
+{
+  "schemaVersion": 1,
+  "title": "Retention cohort readout from event table",
+  "objective": "Measure whether the cohort-analysis craftbook defines retention, emits a cohort matrix, and explains trend/incomplete cohorts.",
+  "tags": [
+    "data"
+  ],
+  "prompt": "In the `Cohort Analysis Eval` project, read `data/events.csv`, define weekly signup cohorts and week-retained activity, write `data/cohorts.json`, and write `tasks/eval/readout.md` with a retention table, W1/W2 trend, best/worst cohorts, incomplete cohort caveat, and one next action. Use unique users per signup_week as cohort size. Period WN means active exactly N weeks after signup; missing activity in an observed week counts as 0, but future unobserved periods are null. Treat 2026-W23 as the latest observed activity week; therefore 2026-W22 W2 would fall in 2026-W24 and must be marked null/incomplete, not estimated. `data/cohorts.json` must be valid JSON only, with no Markdown, comments, formulas, or ratio strings, using this shape: `{ \"cohorts\": [{ \"signup_week\": \"2026-W20\", \"total_users\": 3, \"retention\": { \"W0\": 1, \"W1\": 0.67, \"W2\": 0.33 } }] }`. Use numeric decimal rates rounded to two decimals, or null for incomplete periods.",
+  "setup": {
+    "projectName": "Cohort Analysis Eval",
+    "about": "Self-contained eval project for cohort-analysis. User activity events are in workspace/data/events.csv.",
+    "missionObjectives": "Compute and explain cohort retention with precise definitions, JSON matrix, trend, caveats, and next action.",
+    "files": [
+      {
+        "path": "data/events.csv",
+        "content": "user_id,signup_week,activity_week,event\nu1,2026-W20,2026-W20,core_use\nu1,2026-W20,2026-W21,core_use\nu1,2026-W20,2026-W22,core_use\nu2,2026-W20,2026-W20,core_use\nu2,2026-W20,2026-W21,core_use\nu3,2026-W20,2026-W20,core_use\nu4,2026-W21,2026-W21,core_use\nu4,2026-W21,2026-W22,core_use\nu5,2026-W21,2026-W21,core_use\nu5,2026-W21,2026-W22,core_use\nu6,2026-W21,2026-W21,core_use\nu7,2026-W22,2026-W22,core_use\nu7,2026-W22,2026-W23,core_use\nu8,2026-W22,2026-W22,core_use\nu9,2026-W22,2026-W22,core_use\nu9,2026-W22,2026-W23,core_use\n"
+      }
+    ],
+    "worker": {
+      "name": "Joris",
+      "role": "Data Analyst"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "mocks": [],
+  "success": {
+    "summary": "workspace/readout.md and data/cohorts.json define retention, show cohort rows/periods, report trend, flag incomplete cohorts, and recommend an action.",
+    "deliverables": [
+      {
+        "path": "tasks/eval/readout.md",
+        "kind": "markdown-report",
+        "minBytes": 1500,
+        "checks": [
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "cohort|retention|W0|W1|W2",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "2026-W20|2026-W21|2026-W22",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "100%|67%|66%|33%|0%|trend",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "incomplete|recent cohort|caveat|next action",
+            "flags": "i"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "2026-W22[\\s\\S]{0,250}(incomplete|null|not yet observed|2026-W24)",
+            "flags": "i",
+            "label": "mark W22 W2 as incomplete instead of estimating it"
+          },
+          {
+            "kind": "contains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "2026-W21[^\\n]*(?:0%|0\\.00)",
+            "flags": "i",
+            "label": "readout must show 2026-W21 W2 as observed zero retention"
+          },
+          {
+            "kind": "notContains",
+            "file": "tasks/eval/readout.md",
+            "pattern": "37\\.5%|\\b4\\s+users\\b|total users\\s*\\|\\s*4|2026-W21[^\\n]*(?:N/A|null|incomplete)|W2 for 2026-W21[^\\n]*(?:N/A|null|incomplete)",
+            "flags": "i",
+            "label": "do not invent a fourth 2026-W22 user, estimate W2, or mark observed W21 W2 as incomplete"
+          }
+        ]
+      },
+      {
+        "path": "data/cohorts.json",
+        "kind": "json",
+        "minBytes": 250,
+        "checks": [
+          {
+            "kind": "sniff",
+            "file": "data/cohorts.json",
+            "sniff": "json-valid"
+          },
+          {
+            "kind": "contains",
+            "file": "data/cohorts.json",
+            "pattern": "2026-W20|2026-W21|2026-W22|W0|W1",
+            "flags": "i"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[0].signup_week",
+            "value": "2026-W20",
+            "label": "first cohort must be 2026-W20"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[0].total_users",
+            "value": 3,
+            "label": "2026-W20 has exactly 3 unique signup users"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[0].retention.W2",
+            "value": 0.33,
+            "label": "2026-W20 W2 retention is 1/3 rounded to 0.33"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[0].retention.W1",
+            "value": 0.67,
+            "label": "2026-W20 W1 retention is 2/3 rounded to 0.67"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[1].signup_week",
+            "value": "2026-W21",
+            "label": "second cohort must be 2026-W21"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[1].total_users",
+            "value": 3,
+            "label": "2026-W21 has exactly 3 unique signup users"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[1].retention.W1",
+            "value": 0.67,
+            "label": "2026-W21 W1 retention is 2/3 because u4 and u5 were active in W22"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[1].retention.W2",
+            "value": 0,
+            "label": "2026-W21 W2 retention is 0/3 because no W21 users were active in W23"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[2].signup_week",
+            "value": "2026-W22",
+            "label": "third cohort must be 2026-W22"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[2].total_users",
+            "value": 3,
+            "label": "2026-W22 has exactly 3 unique signup users"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[2].retention.W1",
+            "value": 0.67,
+            "label": "2026-W22 W1 retention is 2/3 rounded to 0.67"
+          },
+          {
+            "kind": "jsonPathEquals",
+            "file": "data/cohorts.json",
+            "path": "cohorts[2].retention.W2",
+            "value": null,
+            "label": "2026-W22 W2 retention is incomplete/null because W24 is outside the observed window"
+          },
+          {
+            "kind": "notContains",
+            "file": "data/cohorts.json",
+            "pattern": "37\\.5|0\\.375|\"total_users\"\\s*:\\s*4",
+            "flags": "i",
+            "label": "do not invent a fourth 2026-W22 user or estimate W2"
+          }
+        ]
+      }
+    ]
+  },
+  "rubric": {
+    "artifact": {
+      "path": "tasks/eval/readout.md",
+      "kind": "markdown"
+    },
+    "axes": [
+      {
+        "name": "accuracy",
+        "description": "Computed figures are correct against the seeded records."
+      },
+      {
+        "name": "grounding",
+        "description": "Every claim traces to the seeded data — nothing invented."
+      },
+      {
+        "name": "structure",
+        "description": "Tables and sections make the analysis checkable at a glance."
+      },
+      {
+        "name": "actionability",
+        "description": "Recommendations follow from the findings and are concretely doable."
+      }
+    ]
+  },
+  "qualityFocus": [
+    "definition before math",
+    "JSON retention matrix",
+    "trend readout"
+  ]
+}

package/data/craftbook-templates/co/company-research-brief/manifest.json

@@ -3,6 +3,7 @@
   "kind": "craftbook-template",
   "id": "company-research-brief",
   "role": "general",
+  "category": "business",
   "name": "Company Research Brief",
   "description": "A sharp, dated, sourced brief on a company before an interview or application — what they do and how they make money, recent signals, team and role context, culture as evidenced, likely interview themes, and five specific questions to ask THEM. Skimmable in ten minutes; every claim dated or marked UNKNOWN rather than guessed. Use before any serious interview or when deciding whether to apply.",
   "tags": [

package/data/craftbook-templates/co/company-research-brief/versions/1.0.2/craftbook.json

@@ -0,0 +1,151 @@
+{
+  "id": "company-research-brief",
+  "name": "Company Research Brief",
+  "description": "A sharp, dated, sourced brief on a company before an interview or application — what they do and how they make money, recent signals, team and role context, culture as evidenced, likely interview themes, and five specific questions to ask THEM. Skimmable in ten minutes; every claim dated or marked UNKNOWN rather than guessed. Use before any serious interview or when deciding whether to apply.\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. Lock the brief template (planner) — sections, sourcing rules, and what a ten-minute skim must deliver → gated on artifact `{{workPath}}/brief-scope.md` (markdown-notes)\n2. Gather the facts (researcher) — dated, sourced notes keyed to the template sections → gated on artifact `{{workPath}}/research-notes.md` (markdown-notes)\n3. Write the brief (copywriter) — the ten-minute brief, led by what matters for the interview → gated on artifact `{{workPath}}/reports/company-brief.md` (markdown-report)\n\nThe gates never advance with an unmet criterion, and loop back to the\nowning phase to fix named gaps.\n",
+  "entryStepId": "scope",
+  "triggers": [
+    "research this company",
+    "company brief before my interview",
+    "prep me on this company",
+    "what should I know about this company"
+  ],
+  "steps": [
+    {
+      "id": "scope",
+      "name": "Lock the brief template",
+      "description": "sections, sourcing rules, and what a ten-minute skim must deliver",
+      "prompt": "Lock the shape before researching. Step 1: confirm the target company, role, and interview round from the project record or the user (briefly). Step 2: lock the brief's section template: what they do + business model; recent news and signals (each dated); team and role context; culture as evidenced by concrete facts (not the careers-page adjectives); likely interview themes for this role; five sharp questions to ask THEM. Step 3: set the sourcing rules — every claim carries a date or a source, anything unverifiable is marked UNKNOWN, nothing is guessed. Step 4: set the size budget: skimmable in ten minutes. Step 5: write an acceptance-criteria checklist ('every section present', 'claims dated/sourced or UNKNOWN', 'questions specific to this company — would not fit a competitor', 'ten-minute skim'). Write the template + checklist via write_task_note AND to {{workPath}}/brief-scope.md.\n\nThe deliverable `{{workPath}}/brief-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}}/brief-scope.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/brief-scope.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/brief-scope.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "scope",
+        "maxAttempts": 3
+      },
+      "next": "research"
+    },
+    {
+      "id": "research",
+      "name": "Gather the facts",
+      "description": "dated, sourced notes keyed to the template sections",
+      "prompt": "Gather facts for every template section. Step 1: use the web research capabilities on your current roster when available; otherwise work from materials in the project and say so at the top of the notes. Step 2: collect facts into structured notes keyed by template section, each with its date and source. Step 3: hunt specifically for recent signals — funding, launches, leadership changes, layoffs, public filings, engineering-blog themes — and for evidence of how the team actually works. Step 4: mark anything you could not verify as UNKNOWN rather than filling the hole with a plausible guess. Step 5: note two or three observations that could become sharp questions to ask them. Write the notes via write_task_note AND to {{workPath}}/research-notes.md.\n\nThe deliverable `{{workPath}}/research-notes.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}}/research-notes.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/research-notes.md",
+            "bytes": 120,
+            "artifact": true
+          },
+          {
+            "kind": "sniff",
+            "file": "{{workPath}}/research-notes.md",
+            "sniff": "nonempty",
+            "artifact": true
+          }
+        ],
+        "onReject": "research",
+        "maxAttempts": 3
+      },
+      "next": "write"
+    },
+    {
+      "id": "write",
+      "name": "Write the brief",
+      "description": "the ten-minute brief, led by what matters for the interview",
+      "prompt": "Assemble {{workPath}}/reports/company-brief.md to the locked template. Step 1: lead with the two or three things that matter most for THIS interview — the reader may only get the first screen. Step 2: fill every template section from the research notes; every claim traces to a dated note, UNKNOWNs stay visibly UNKNOWN. Step 3: write the likely-interview-themes section as preparation the candidate can act on tonight. Step 4: close with the five questions to ask them — each specific enough that it would make no sense asked of a competitor. Step 5: trim to the ten-minute skim budget; cut color, keep signal. On a loop-back, fix ONLY the named issues.\n\nThe deliverable `{{workPath}}/reports/company-brief.md` lands in the project's artifacts drawer — write it with `write_artifact` and read it back with `read_artifact`; the shipped workspace stays untouched.",
+      "suggestedRole": "copywriter",
+      "advanceWhen": {
+        "file": "{{workPath}}/reports/company-brief.md",
+        "minBytes": 1,
+        "sniff": "nonempty",
+        "artifact": true
+      },
+      "gate": {
+        "at": "completion",
+        "checks": [
+          {
+            "kind": "minBytes",
+            "file": "{{workPath}}/reports/company-brief.md",
+            "bytes": 1500,
+            "artifact": true
+          },
+          {
+            "kind": "contains",
+            "file": "{{workPath}}/reports/company-brief.md",
+            "pattern": "(?:^|\\n)#{1,3}\\s+\\S",
+            "flags": "i",
+            "label": "at least one markdown heading",
+            "artifact": true
+          }
+        ],
+        "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 the brief. Step 1: check every template section is present and filled from the notes — no section silently dropped. Step 2: spot-check claims against the research notes for dating/sourcing; FAIL any undated claim that isn't marked UNKNOWN. Step 3: read the five questions as the interviewer would — FAIL any that could be asked of any company. Step 4: time the skim — if it reads past ten minutes, FAIL with what to cut. Write PASS/FAIL per criterion.\n\nThe deliverable lives in the project's artifacts drawer — open `{{workPath}}/reports/company-brief.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: \"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": "{{workPath}}/reports/company-brief.md",
+          "artifact": true
+        }
+      ],
+      "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/co/company-research-brief/versions/1.0.2/test.json

@@ -0,0 +1,100 @@
+{
+  "schemaVersion": 1,
+  "title": "Company Research Brief smoke eval",
+  "objective": "Self-contained smoke eval for the Company Research Brief 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": "Company Research Brief Eval",
+    "about": "Self-contained eval project for company-research-brief. Seeded inputs are under workspace/source or workspace/fixtures; final deliverable is workspace/index.html.",
+    "missionObjectives": "Use the Company Research Brief 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": "# Company Research Brief 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: company-research-brief - Company Research Brief.\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"
+    },
+    "craftbookParams": {
+      "workPath": "tasks/eval"
+    }
+  },
+  "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/co/comparison-matrix/manifest.json

@@ -3,6 +3,7 @@
   "kind": "craftbook-template",
   "id": "comparison-matrix",
   "role": "general",
+  "category": "research",
   "name": "Option Comparison Matrix",
   "description": "Compare a set of options against weighted criteria and ship an interactive HTML comparison matrix with a recommendation: a planner locks the options, the criteria, and their weights, a researcher fills every cell with an evidence-backed value, and a developer builds a single-file HTML page with a sortable/filterable comparison table and a scored recommendation. The ordering matters because criteria and weights chosen after seeing the data bias the winner, and a static table buries the decision; locking weighted criteria first makes the comparison fair and lets the build compute a transparent score. Use this for option/tool/vendor comparisons, decision matrices, buying guides, 'help me choose between X, Y, Z', and weighted pros/cons tables.",
   "tags": [