@lavralabs/lavra 0.7.0

package/plugins/lavra/gemini/commands/lavra-qa.toml

@@ -0,0 +1,360 @@
+# Generated by lavra v0.6.0
+# DO NOT EDIT - changes will be overwritten on next install
+
+description = "Browser-based QA verification of the running app -- systematic testing from the user's perspective"
+
+prompt = """
+
+<objective>
+Verify that implemented changes work correctly from the user's perspective by running systematic browser-based tests against the running application. Sits between /lavra-work (implementation) and shipping, catching visual regressions, broken interactions, console errors, and workflow breakages that unit tests miss.
+</objective>
+
+<execution_context>
+<qa_target> {{args}} </qa_target>
+</execution_context>
+
+<guardrails>
+
+**DO NOT use Chrome MCP tools (mcp__claude-in-chrome__*).**
+
+This command uses the `agent-browser` CLI exclusively. The agent-browser CLI is a Bash-based tool from Vercel that runs headless Chromium. It is NOT the same as Chrome browser automation via MCP.
+
+If you find yourself calling `mcp__claude-in-chrome__*` tools, STOP. Use `agent-browser` Bash commands instead.
+
+**DO NOT force browser QA on non-UI work.** If the diff shows only backend/CLI/library/infra changes with no web UI impact, say so and suggest skipping. Do not waste time opening a browser when there is nothing visual to test.
+
+</guardrails>
+
+<process>
+
+### Phase 0: Mode Detection
+
+Parse arguments:
+
+- `--quick` flag: smoke test mode (load pages, check for errors, done)
+- No flag: full mode (all test scenarios, interactive elements, edge cases)
+- Bead ID: use bead description to understand what was implemented and what to verify
+
+If a bead ID is provided:
+```bash
+bd show {BEAD_ID} --json
+```
+
+Read the bead description to understand acceptance criteria and what the implementation should do. This informs what to test.
+
+### Phase 1: Scope Detection
+
+**Identify changed files:**
+
+```bash
+# If on a feature branch
+git diff --name-only $(git merge-base HEAD main)..HEAD
+
+# Fallback: unstaged + staged changes
+git diff --name-only HEAD
+```
+
+**Detect framework and map files to routes:**
+
+| Framework | Detection | Route Mapping |
+|-----------|-----------|---------------|
+| Next.js | `next.config.*` or `src/app/` | `src/app/**/page.tsx` -> URL path |
+| Rails | `Gemfile` with `rails` | `config/routes.rb` + changed controllers/views |
+| Django | `manage.py` or `urls.py` | `urls.py` patterns + changed views/templates |
+| Laravel | `artisan` | `routes/web.php` + changed controllers/views |
+| Remix | `remix.config.*` | `app/routes/` directory structure |
+| SvelteKit | `svelte.config.*` | `src/routes/` directory structure |
+| Nuxt | `nuxt.config.*` | `pages/` directory structure |
+| Generic SPA | `index.html` + router config | Router config file |
+
+Read the relevant routing file to map changed files to URLs:
+
+```bash
+# Rails example
+cat config/routes.rb
+# Next.js example
+find src/app -name "page.tsx" -o -name "page.js" | head -20
+# Django example
+cat */urls.py
+```
+
+**Check for non-UI changes:**
+
+If ALL changed files match these patterns, suggest skipping QA:
+- `*.rb` models/services/jobs with no view/controller changes
+- `*.py` without template/view changes
+- API-only endpoints (serializers, API controllers)
+- Database migrations only
+- CLI tools, libraries, gems, packages
+- Infrastructure (Dockerfile, CI configs, terraform)
+- Documentation only
+
+If non-UI detected, present:
+
+> "Changes appear to be backend/infrastructure only with no UI impact. Browser QA would not add value here. Skip QA?"
+
+Proceed only if user confirms there IS a UI to test.
+
+**Build the test URL list** from the file-to-route mapping. Present it to the user:
+
+```markdown
+## QA Test Plan
+
+**Framework detected:** [framework]
+**Changed routes:**
+
+| Route | Changed Files | What to Verify |
+|-------|--------------|----------------|
+| /users | users_controller.rb, index.html.erb | User listing renders correctly |
+| /settings | settings.js, settings.css | Settings page layout and interactions |
+```
+
+Use **AskUserQuestion tool**:
+
+**Question:** "Here is the QA test plan. What is the base URL for the running app?"
+
+**Options:**
+1. **http://localhost:3000** (default)
+2. **http://localhost:5173** (Vite default)
+3. **http://localhost:8000** (Django/Laravel default)
+4. **Custom URL** - I will provide it
+
+Also ask if they want to add or remove any routes from the test plan.
+
+### Phase 2: Server Verification
+
+**Verify agent-browser is installed:**
+
+```bash
+command -v agent-browser >/dev/null 2>&1 && echo "Ready" || (echo "Installing..." && npm install -g agent-browser && agent-browser install)
+```
+
+If installation fails, inform the user and stop.
+
+**Ask browser mode:**
+
+Use **AskUserQuestion tool**:
+
+**Question:** "Do you want to watch the browser tests run?"
+
+**Options:**
+1. **Headed (watch)** - Opens visible browser window so you can see tests run
+2. **Headless (faster)** - Runs in background, faster but invisible
+
+Store the choice and use `--headed` flag when user selects "Headed".
+
+**Verify server is reachable:**
+
+```bash
+agent-browser open {BASE_URL}
+agent-browser snapshot -i
+```
+
+If server is not running:
+
+> "Server is not reachable at {BASE_URL}. Please start your development server and confirm, or provide the correct URL."
+
+Do not proceed until the server responds.
+
+### Phase 3: Test Plan Generation
+
+For each affected route, generate test scenarios based on mode:
+
+**--quick mode (smoke test):**
+- Page loads without errors
+- No console errors/warnings
+- Key heading/content is present
+- Screenshot for evidence
+
+**Full mode (default):**
+- Page loads without errors
+- Console has no errors/warnings
+- Key headings and content render correctly
+- Navigation elements work (links, tabs, breadcrumbs)
+- Forms: fields present, validation fires, submission works
+- Buttons and interactive elements respond to clicks
+- Changed functionality behaves as specified in the bead
+- Data displays correctly (tables, lists, cards)
+- Responsive check: viewport resize if layout changes were made
+- Authentication-gated pages accessible when logged in
+
+Present the test plan for user approval before executing.
+
+### Phase 4: Execution
+
+For each route in the test plan:
+
+**Step 1: Navigate and assess**
+```bash
+agent-browser open "{BASE_URL}{route}"
+agent-browser snapshot -i
+agent-browser get title
+```
+
+**Step 2: Check for errors**
+```bash
+agent-browser snapshot -i --json
+```
+
+Look for error messages, 404/500 pages, missing content, broken layouts in the snapshot output.
+
+**Step 3: Test interactive elements (full mode only)**
+
+For forms:
+```bash
+agent-browser snapshot -i
+# Identify form fields from snapshot refs
+agent-browser fill @e1 "test input"
+agent-browser click @submit_ref
+agent-browser snapshot -i  # Check result
+```
+
+For navigation:
+```bash
+agent-browser click @nav_ref
+agent-browser snapshot -i  # Verify navigation worked
+agent-browser back
+```
+
+For dynamic content:
+```bash
+agent-browser click @trigger_ref
+agent-browser wait 1000
+agent-browser snapshot -i  # Check updated state
+```
+
+**Step 4: Take screenshots**
+```bash
+agent-browser screenshot qa-{route-slug}.png
+agent-browser screenshot --full qa-{route-slug}-full.png
+```
+
+**Step 5: Record result**
+
+Assign each page a health score:
+- **PASS** - Page loads, no errors, interactions work as expected
+- **WARN** - Page loads but has minor issues (non-critical console warnings, minor visual issues)
+- **FAIL** - Page broken, console errors, interactions fail, content missing
+
+### Phase 5: Handle Failures
+
+When a test fails:
+
+1. **Document the failure:**
+   ```bash
+   agent-browser screenshot qa-fail-{route-slug}.png
+   ```
+
+2. **Ask user how to proceed:**
+
+   Use **AskUserQuestion tool**:
+
+   **Question:** "QA failure on {route}: {description}. How to proceed?"
+
+   **Options:**
+   1. **Fix now** - Investigate and fix the issue
+   2. **Create bead** - Track as a bug for later
+   3. **Skip** - Accept and continue testing
+
+3. **If "Fix now":**
+   - Investigate the root cause
+   - Propose and apply a fix
+   - Re-run the failing test to verify
+
+4. **If "Create bead":**
+   ```bash
+   bd create "QA failure: {description} on {route}" --type bug --priority 1
+   ```
+   Continue testing remaining routes.
+
+5. **If "Skip":**
+   - Log as skipped with reason
+   - Continue testing
+
+### Phase 6: Results
+
+After all routes tested, present the summary:
+
+```markdown
+## QA Results
+
+**Mode:** [quick/full]
+**Base URL:** {BASE_URL}
+**Bead:** {BEAD_ID} (if provided)
+
+### Pages Tested: [count]
+
+| Route | Health | Notes |
+|-------|--------|-------|
+| /users | PASS | |
+| /settings | WARN | Minor layout shift on mobile viewport |
+| /dashboard | FAIL | Console error: TypeError in chart.js |
+
+### Console Errors: [count]
+- [List errors with route where found]
+
+### Failures: [count]
+- {route} - {issue description}
+
+### Beads Created: [count]
+- {BEAD_ID}: {title}
+
+### Screenshots
+- qa-users.png
+- qa-settings.png
+- qa-fail-dashboard.png
+
+### Result: [PASS / WARN / FAIL]
+```
+
+**Log knowledge for unexpected findings:**
+
+```bash
+bd comments add {BEAD_ID} "LEARNED: {unexpected behavior discovered during QA}"
+```
+
+**Close the browser:**
+```bash
+agent-browser close
+```
+
+### Phase 7: Next Steps
+
+Use **AskUserQuestion tool**:
+
+**Question:** "QA complete. Result: {PASS/WARN/FAIL}. What next?"
+
+**Options (if PASS):**
+1. **Run `/lavra-review`** - Code review before shipping
+2. **Close bead** - Mark as complete: `bd close {BEAD_ID}`
+3. **Ship it** - Push and create PR
+
+**Options (if WARN or FAIL):**
+1. **Fix issues** - Address failures before shipping
+2. **Run `/lavra-review`** - Code review (issues noted but accepted)
+3. **Create beads for failures** - Track issues separately and ship
+4. **Re-run QA** - Test again after fixes
+
+</process>
+
+<success_criteria>
+- [ ] Changed files identified and mapped to routes
+- [ ] Non-UI changes correctly detected (skip suggested when appropriate)
+- [ ] Dev server verified as running
+- [ ] All affected pages tested with agent-browser CLI
+- [ ] Each page has a PASS/WARN/FAIL health score
+- [ ] Console errors captured and reported
+- [ ] Screenshots taken as evidence
+- [ ] Failures documented with reproduction steps
+- [ ] Fix beads created for unresolved failures
+- [ ] Knowledge logged for unexpected behaviors
+</success_criteria>
+
+<handoff>
+After QA completes:
+1. **Run `/lavra-review`** - Multi-agent code review
+2. **Fix failures** - Address any FAIL results
+3. **Ship** - Push changes and create PR
+</handoff>
+
+"""

package/plugins/lavra/gemini/commands/lavra-quick.toml

@@ -0,0 +1,181 @@
+# Generated by lavra v0.6.0
+# DO NOT EDIT - changes will be overwritten on next install
+
+description = "Fast-track small tasks — abbreviated plan then straight to execution"
+
+prompt = """
+
+<objective>
+Fast-track small tasks with an abbreviated plan and immediate execution. Skips brainstorm and deepen phases, runs a MINIMAL plan (1-3 child tasks), then transitions directly to `/lavra-work`. Still captures knowledge throughout.
+</objective>
+
+<execution_context>
+<raw_argument> #{{args}} </raw_argument>
+
+**Determine if the argument is a bead ID or a task description:**
+
+Check if the argument matches a bead ID pattern:
+- Pattern: lowercase alphanumeric segments separated by hyphens (e.g., `fix-auth`, `beads-123`)
+- Regex: `^[a-z0-9]+-[a-z0-9]+(-[a-z0-9]+)*$`
+
+**If bead ID pattern:**
+
+1. Load the bead:
+   ```bash
+   bd show "#$ARGUMENTS" --json
+   ```
+2. If it exists: extract title and description, announce "Quick-tracking bead #{{args}}: {title}"
+3. If not found: report error and stop
+
+**If task description:**
+- Create a bead:
+  ```bash
+  bd create --title="{concise title}" --description="#$ARGUMENTS" --type=task
+  ```
+- Capture the new bead ID
+
+**If empty:**
+- Ask: "What small task do you want to quick-track? Provide a bead ID or describe the task."
+- Do not proceed until input is provided
+</execution_context>
+
+<context>
+Use `/lavra-quick` for small, well-understood tasks: bug fixes, config changes, small refactors, adding a field, writing a utility function.
+
+If the task turns out larger than expected, the scope escalation check (step 3) will catch it and offer to switch to `/lavra-design`.
+</context>
+
+<process>
+
+### 1. Quick Context Scan
+
+Run in parallel:
+
+```bash
+# Recall relevant knowledge
+.lavra/memory/recall.sh "{keywords from task}"
+```
+
+```bash
+# Quick repo scan for related patterns
+# (grep/glob for relevant files based on task description)
+```
+
+Output recall results before continuing. If nothing found, state "No relevant knowledge found."
+
+### 2. Abbreviated Plan (MINIMAL)
+
+Create 1-3 child tasks as beads. No deep research, no deepen, no review.
+
+```bash
+bd create "{step title}" --parent {BEAD_ID} -d "## What
+{what to implement}
+
+## Validation
+- [ ] {acceptance criterion}"
+```
+
+Keep descriptions short -- this is the fast path. Each child bead needs only What and Validation sections.
+
+If two tasks touch the same file, add a dependency:
+```bash
+bd dep add {later_bead} {earlier_bead}
+```
+
+### 3. Scope Escalation Check
+
+After creating the abbreviated plan but BEFORE starting implementation, evaluate whether the task has outgrown quick-fix territory. Check for these signals:
+
+- **File count**: More than 3 files need changes
+- **Cross-bead dependencies**: Dependencies on other existing beads discovered
+- **Architectural decisions**: The task requires architectural choices, not just implementation choices
+- **Security implications**: Auth, permissions, data exposure, or input validation concerns found
+- **Multi-component impact**: Changes span multiple components, services, or layers
+- **Change volume**: Estimated total changes exceed ~100 lines
+
+**If one or more signals are detected**, pause execution and report:
+
+<scope_escalation>
+"This task has grown beyond quick-fix scope.
+
+Signals detected:
+- {list each signal that fired with a brief explanation}
+
+Switch to /lavra-design for proper planning? This preserves all work done so far."
+</scope_escalation>
+
+**If the user accepts escalation:**
+
+1. Save current progress -- update the parent bead description with a note summarizing the abbreviated plan and the escalation signals:
+   ```bash
+   bd comments add {BEAD_ID} "DECISION: Escalated from /lavra-quick to /lavra-design. Signals: {signals}. Child tasks preserved as starting point."
+   ```
+2. Invoke `/lavra-design` with the bead ID so the full planning pipeline picks up where this left off.
+3. Stop the lavra-quick workflow. Do not continue to step 4.
+
+**If the user declines escalation:**
+
+1. Log the decision:
+   ```bash
+   bd comments add {BEAD_ID} "DECISION: User chose to proceed with /lavra-quick despite scope signals: {signals}. Rationale: user preference."
+   ```
+2. Continue to step 4.
+
+**If no signals are detected**, proceed to step 4 without interruption.
+
+### 4. Begin Execution
+
+Update the parent bead status and transition to execution:
+
+```bash
+bd update {BEAD_ID} --status in_progress
+```
+
+**Light deviation rules for quick tasks:**
+- Auto-fix bugs and blockers that prevent task completion -> log `DEVIATION:`
+- Do NOT expand scope beyond the original task description
+- If you encounter something that requires scope expansion, log it and move on
+
+Execute using the `/lavra-work` workflow on the first ready child bead. Follow all `/lavra-work` phases (Quick Start, Execute, Quality Check, Ship It) -- the abbreviated plan does not mean abbreviated execution.
+
+**Log knowledge as you work** -- at least one LEARNED/DECISION/FACT/PATTERN/DEVIATION comment per task:
+
+```bash
+bd comments add {BEAD_ID} "LEARNED: {insight}"
+bd comments add {BEAD_ID} "DEVIATION: {what was changed outside original scope and why}"
+```
+
+### 5. Wrap Up
+
+After all child tasks are complete:
+
+1. Run tests and linting
+2. Commit with conventional format
+3. Close the bead: `bd close {BEAD_ID}`
+
+</process>
+
+<success_criteria>
+- Bead created (if description provided) or loaded (if ID provided)
+- 1-3 child tasks created with What/Validation sections
+- All tasks executed and tests passing
+- At least one knowledge comment captured
+- Bead closed on completion
+</success_criteria>
+
+<guardrails>
+- Do NOT use for complex features, architectural changes, or tasks with unclear requirements
+- If scope creep is detected, the formal escalation check in step 3 handles it -- do not skip that checkpoint
+- Do NOT skip knowledge capture -- the fast path still feeds the memory system
+- Do NOT skip tests -- abbreviated planning does not mean lower quality
+</guardrails>
+
+<handoff>
+After completion, present options:
+
+1. **Quick-track another task** -- run `/lavra-quick` again
+2. **Review the work** -- run `/lavra-review` for a code review
+3. **Checkpoint** -- run `/lavra-checkpoint` to save progress
+</handoff>
+
+"""