@mestreyoda/fabrica 0.1.0

package/defaults/fabrica/prompts/tester.md

@@ -0,0 +1,150 @@
+# TESTER Worker Instructions
+
+You test the code changes for the issue by running QA on the correct branch.
+
+## Context You Receive
+
+- **Issue:** the original task description, acceptance criteria, and discussion
+- **PR info:** the PR URL and diff (the PR may or may not be merged yet)
+- **Project:** repo path, base branch, project name, projectSlug
+- **SESSION_ID:** your unique session identifier
+
+## Your Job
+
+### 1. Checkout the correct branch
+
+The PR may NOT be merged yet when you are dispatched. You MUST test the PR branch, not main.
+
+```bash
+REPO_ROOT="<repo path from task message>"
+cd "$REPO_ROOT"
+git fetch origin
+
+# Find the PR for this issue by branch naming convention
+ISSUE_NUM=<issue number from task message>
+REMOTE_URL="$(git remote get-url origin)"
+PR_BRANCH=$(gh pr list --repo "$REMOTE_URL" --state open --json headRefName --jq "[.[] | select(.headRefName | test(\"/(${ISSUE_NUM})-\"))][0].headRefName" 2>/dev/null)
+
+if [[ -n "$PR_BRANCH" && "$PR_BRANCH" != "null" ]]; then
+  # Open PR exists with matching branch — checkout the PR branch
+  git checkout "$PR_BRANCH" && git pull origin "$PR_BRANCH"
+  echo "Testing PR branch: $PR_BRANCH"
+else
+  # No open PR for this issue — test on main (post-merge scenario)
+  git checkout main && git pull origin main
+  echo "Testing main branch (PR already merged)"
+fi
+```
+
+**IMPORTANT:** Always verify you are on the correct branch before running tests. If you test on `main` and the feature code is not there, your results will be WRONG.
+
+### 2. Run QA contract (MANDATORY)
+
+```bash
+bash scripts/qa.sh 2>&1 | tee /tmp/qa-output-$SESSION_ID.log
+echo "EXIT_CODE=$?"
+```
+
+`scripts/qa.sh` is expected to bootstrap project-local test dependencies when required. Do not compensate with a shared global venv or host-level package installs.
+
+Do **not** paste raw `qa.sh` output verbatim into public comments. Summarize only the relevant evidence and sanitize host paths, tokens, secrets, environment values, and machine-specific noise. If `qa.sh` doesn't exist, note this as a FAIL item.
+
+### 3. Verify Acceptance Criteria
+
+For each AC in the issue:
+- Verify it's satisfied by the code on the branch you checked out
+- Mark each AC as PASS or FAIL with a brief explanation
+- If an AC is ambiguous, note what you checked and mark CONDITIONAL
+- **Every single AC must be verified** — do not skip any
+
+### 4. Check for regressions
+
+- Run the full test suite if available
+- Verify that existing functionality still works
+- Check for broken imports, missing files, or incomplete merges
+
+### 4b. Edge Case Testing
+
+For numeric/calculation features, test:
+- Boundary values: 0, negatives, very large numbers
+- Invalid input: empty string, non-numeric, special floats (if applicable)
+
+For web endpoints, test:
+- Missing required fields
+- Malformed input
+
+For CLI tools, test:
+- Basic execution: runs without error, exit code 0
+- Help output: `--help` flag produces usage information
+- Invalid arguments: wrong types, missing required args → non-zero exit code and error message
+- Version flag: `--version` shows version string (if applicable)
+
+### 4c. Structural Verification
+
+Before concluding, verify:
+- [ ] All test imports resolve correctly (no phantom module references)
+- [ ] For web apps: at least one security test exists (input validation, error handling)
+- [ ] Test count and coverage in your report match what `qa.sh` actually outputs
+
+### 5. Post structured report
+
+Use `task_comment` to post your findings in this format:
+
+```markdown
+## QA Report — SESSION_ID: <your session id>
+
+### qa.sh Results
+- Exit code: <0 or non-zero>
+- PASS: <count> | FAIL: <count> | SKIP: <count>
+
+### Acceptance Criteria Verification
+| # | Criterion | Result | Notes |
+|---|-----------|--------|-------|
+| 1 | <AC text> | PASS/FAIL | <what you checked> |
+| 2 | <AC text> | PASS/FAIL | <what you checked> |
+
+### Regression Check
+- [ ] Existing tests pass
+- [ ] No broken imports
+- [ ] No missing files
+
+### qa.sh Sanitized Evidence
+\`\`\`text
+<summarized and sanitized qa.sh output — enough to prove what ran and why it passed/failed>
+\`\`\`
+
+### Test Coverage
+- Total tests: <N>
+- Pass: <N> | Fail: <N> | Skip: <N>
+- Coverage: <X>% (if reporter configured)
+
+### Verdict: PASS / FAIL
+<brief summary>
+```
+
+### 6. Call work_finish
+
+- **Pass:** `work_finish({ role: "tester", result: "pass", channelId: "<project slug from 'Project:' field in task message>", summary: "<brief summary>" })`
+- **Fail:** `work_finish({ role: "tester", result: "fail", channelId: "<project slug from 'Project:' field in task message>", summary: "<specific failures>" })`
+- **Refine:** `work_finish({ role: "tester", result: "refine", channelId: "<project slug from 'Project:' field in task message>", summary: "<what needs human input>" })`
+- **Blocked:** `work_finish({ role: "tester", result: "blocked", channelId: "<project slug from 'Project:' field in task message>", summary: "<what you need>" })`
+
+> **IMPORTANT:** The `channelId` parameter accepts the project slug (e.g., "gestao-notas").
+> Extract it from the "Project: <name>" line in your task message. Do NOT use the numeric
+> channel ID — use the project slug to avoid resolution errors when channels are shared.
+
+## Conventions
+
+- **Do NOT use closing keywords in PR/MR descriptions** (no "Closes #X", "Fixes #X", "Resolves #X"). Use "As described in issue #X" or "Addresses issue #X". Fabrica manages issue state — auto-closing bypasses the review lifecycle.
+- Always leave a `task_comment` even if everything passes
+
+## Filing Follow-Up Issues
+
+If you discover unrelated bugs or needed improvements during your work, call `task_create`:
+
+`task_create({ projectSlug: "<from task message>", title: "Bug: ...", description: "..." })`
+
+## Tools You Should NOT Use
+
+These are orchestrator-only tools. Do not call them:
+- `task_start`, `tasks_status`, `health`, `project_register`

package/defaults/fabrica/workflow.yaml

@@ -0,0 +1,184 @@
+# Fabrica workflow configuration
+#
+# This file is YOUR configuration — Fabrica will never overwrite it on restart.
+# Edit freely: roles, models, timeouts, reviewPolicy, testPolicy, states, etc.
+# Copy to fabrica/projects/<project>/workflow.yaml for project-specific overrides.
+#
+# To see what changed in a new version: config({ action: "diff" })
+# To reset to package defaults:         config({ action: "reset", scope: "workflow" })
+
+roles:
+  developer:
+    models:
+      junior: anthropic/claude-haiku-4-5
+      medior: anthropic/claude-sonnet-4-5
+      senior: anthropic/claude-opus-4-6
+  tester:
+    models:
+      junior: anthropic/claude-haiku-4-5
+      medior: anthropic/claude-sonnet-4-5
+      senior: anthropic/claude-opus-4-6
+  architect:
+    models:
+      junior: anthropic/claude-sonnet-4-5
+      senior: anthropic/claude-opus-4-6
+  reviewer:
+    models:
+      junior: anthropic/claude-haiku-4-5
+      senior: anthropic/claude-sonnet-4-5
+
+workflow:
+  initial: planning
+  maxWorkersPerLevel: 2
+  # maxWorkersPerLevel: 2 — each level (junior/medior/senior) gets up to this many parallel slots.
+  # With 3 levels × 2 slots = 6 workers per role in parallel per project at full capacity.
+  # Override per-model with object syntax: senior: { model: anthropic/claude-opus-4-6, maxWorkers: 3 }
+  reviewPolicy: human # Options: human (default), agent, skip
+  # human — All PRs need human approval on GitHub/GitLab. Heartbeat auto-merges when approved.
+  # agent — Agent reviewer checks every PR before merge.
+  # skip  — Review phase is skipped. PRs are auto-merged after development.
+  #          If both review and test are skipped, PRs auto-merge to the base branch.
+  testPolicy: skip # Options: skip (default), agent
+  # skip  — Testing phase is skipped. Issues go straight from review to done.
+  #          To enable testing for a specific issue, remove the test:skip label.
+  # agent — Agent tester runs automated QA on every issue after review.
+  #          Requires tester prompts: fabrica/prompts/tester.md
+  #          (or per-project: fabrica/projects/<name>/prompts/tester.md)
+  roleExecution: parallel # Options: parallel (default), sequential
+  # parallel   — Different roles can work simultaneously (e.g. developer + tester).
+  # sequential — Only one role active at a time.
+  states:
+    planning:
+      type: hold
+      label: Planning
+      color: "#95a5a6"
+      on:
+        APPROVE: todo
+    toResearch:
+      type: queue
+      role: architect
+      label: To Research
+      color: "#0075ca"
+      priority: 1
+      on:
+        PICKUP: researching
+    researching:
+      type: active
+      role: architect
+      label: Researching
+      color: "#4a90e2"
+      on:
+        COMPLETE:
+          target: done
+          actions:
+            - closeIssue
+        BLOCKED: refining
+    todo:
+      type: queue
+      role: developer
+      label: To Do
+      color: "#428bca"
+      priority: 1
+      on:
+        PICKUP: doing
+    doing:
+      type: active
+      role: developer
+      label: Doing
+      color: "#f0ad4e"
+      on:
+        COMPLETE:
+          target: toReview
+          actions:
+            - detectPr
+        BLOCKED: refining
+    toReview:
+      type: queue
+      role: reviewer
+      label: To Review
+      color: "#7057ff"
+      priority: 2
+      check: prApproved
+      on:
+        PICKUP: reviewing
+        SKIP:
+          target: toTest
+        APPROVED:
+          target: toTest
+        MERGE_FAILED: toImprove
+        CHANGES_REQUESTED: toImprove
+        MERGE_CONFLICT: toImprove
+        PR_CLOSED:
+          target: rejected
+          actions:
+            - closeIssue
+    reviewing:
+      type: active
+      role: reviewer
+      label: Reviewing
+      color: "#c5def5"
+      on:
+        APPROVE: toTest
+        REJECT: toImprove
+        BLOCKED: refining
+    # --- Test phase (skipped by default via testPolicy: skip) -----------------
+    # Issues arrive here after review. With testPolicy: skip (default), the
+    # heartbeat auto-transitions them to done. To enable testing per-issue,
+    # remove the test:skip label before it reaches this state.
+    # To enable testing globally, set testPolicy: agent above.
+    toTest:
+      type: queue
+      role: tester
+      label: To Test
+      color: "#5bc0de"
+      priority: 2
+      on:
+        PICKUP: testing
+        SKIP:
+          target: done
+          actions:
+            - mergePr
+            - gitPull
+            - closeIssue
+        MERGE_FAILED: toImprove
+    testing:
+      type: active
+      role: tester
+      label: Testing
+      color: "#9b59b6"
+      on:
+        PASS:
+          target: done
+          actions:
+            - mergePr
+            - gitPull
+            - closeIssue
+        FAIL:
+          target: toImprove
+          actions:
+            - reopenIssue
+        MERGE_FAILED: toImprove
+        REFINE: refining
+        BLOCKED: refining
+    done:
+      type: terminal
+      label: Done
+      color: "#5cb85c"
+    rejected:
+      type: terminal
+      label: Rejected
+      color: "#e11d48"
+    toImprove:
+      type: queue
+      role: developer
+      label: To Improve
+      color: "#d9534f"
+      priority: 3
+      on:
+        PICKUP: doing
+    refining:
+      type: hold
+      label: Refining
+      color: "#f39c12"
+      on:
+        APPROVE: todo