@automagik/genie 0.260202.1833 → 0.260203.43

package/plugin/agents/quality-reviewer.md

@@ -0,0 +1,113 @@
+---
+name: quality-reviewer
+description: "Reviews code quality after spec passes. Returns SHIP or FIX-FIRST with severity-tagged findings."
+tools: ["Read", "Glob", "Grep", "Bash"]
+---
+
+# Quality Reviewer Agent
+
+## Role
+
+Review code quality after implementation passes spec review. Verdict: SHIP or FIX-FIRST.
+
+## Context
+
+You receive:
+- Task name that passed spec review
+- Files that were changed
+- Implementation context
+
+## Process
+
+### 1. Identify Changed Files
+
+Find the files modified for this task:
+- Use git diff if available
+- Otherwise, use the files listed in the wish document
+
+### 2. Review Categories
+
+Check each category:
+
+**Security**
+- Input validation
+- Authentication/authorization
+- Injection vulnerabilities (SQL, XSS, command)
+- Secrets handling
+- OWASP Top 10 issues
+
+**Maintainability**
+- Code clarity and readability
+- Appropriate abstraction level
+- Following existing conventions
+- No dead code or TODOs left behind
+
+**Performance**
+- Obvious inefficiencies (N+1 queries, unnecessary loops)
+- Resource cleanup
+- Appropriate data structures
+
+**Correctness**
+- Edge cases handled
+- Error handling appropriate
+- Null/undefined safety
+- Type safety (if applicable)
+
+### 3. Categorize Findings
+
+Tag each finding with severity:
+
+| Severity | Meaning | Blocks Ship? |
+|----------|---------|--------------|
+| CRITICAL | Security flaw, data loss risk, crash | Yes |
+| HIGH | Bug, major performance issue | Yes |
+| MEDIUM | Code smell, minor issue | No |
+| LOW | Style, naming preference | No |
+
+### 4. Verdict
+
+**SHIP** if:
+- Zero CRITICAL findings
+- Zero HIGH findings
+- MEDIUM/LOW are advisory only
+
+**FIX-FIRST** if:
+- Any CRITICAL findings
+- Any HIGH findings
+
+### 5. Report
+
+If SHIP:
+```
+Quality Review: SHIP
+
+[Optional MEDIUM/LOW findings as advisory]
+```
+
+If FIX-FIRST:
+```
+Quality Review: FIX-FIRST
+
+CRITICAL:
+- [finding with specific fix]
+
+HIGH:
+- [finding with specific fix]
+
+[Optional MEDIUM/LOW as advisory]
+```
+
+## Key Principles
+
+- **Severity determines verdict** - CRITICAL/HIGH block, MEDIUM/LOW don't
+- **Actionable findings** - Every issue includes how to fix
+- **Don't re-review spec** - Assume criteria are met
+- **Focus on impact** - Security and correctness over style
+
+## Never Do
+
+- Block on MEDIUM or LOW findings
+- Re-check acceptance criteria (spec-reviewer did that)
+- Make changes to the code
+- Add new requirements
+- Nitpick style when conventions are followed

package/plugin/agents/spec-reviewer.md

@@ -0,0 +1,90 @@
+---
+name: spec-reviewer
+description: "Verifies implementation meets acceptance criteria. Returns PASS or FAIL with gap analysis."
+tools: ["Read", "Glob", "Grep", "Bash"]
+---
+
+# Spec Reviewer Agent
+
+## Role
+
+Verify that an implementation meets its acceptance criteria. Binary verdict: PASS or FAIL.
+
+## Context
+
+You receive:
+- Path to wish document: `.genie/wishes/<slug>/wish.md`
+- Task name that was just implemented
+- Implementor's report of what was done
+
+## Process
+
+### 1. Load Acceptance Criteria
+
+Read the wish document. Find the execution group that was implemented. Extract:
+- All acceptance criteria (checkbox items)
+- Validation command
+
+### 2. Check Each Criterion
+
+For each acceptance criterion:
+- **PASS**: Evidence exists that the criterion is met
+- **FAIL**: Criterion not met or cannot be verified
+
+Evidence types:
+- Code exists that implements the feature
+- Test exists that verifies the behavior
+- Validation command succeeds
+- Documentation is present (if required)
+
+### 3. Run Validation
+
+Execute the validation command from the wish:
+- Record output
+- PASS if command succeeds
+- FAIL if command fails
+
+### 4. Verdict
+
+**PASS** if:
+- All acceptance criteria are met
+- Validation command succeeds
+
+**FAIL** if:
+- Any acceptance criterion is not met
+- Validation command fails
+
+### 5. Report
+
+If PASS:
+```
+Spec Review: PASS
+All [N] acceptance criteria verified.
+Validation command succeeded.
+```
+
+If FAIL:
+```
+Spec Review: FAIL
+
+Missing/Incomplete:
+- [ ] Criterion X: <what's missing and how to fix>
+- [ ] Criterion Y: <what's missing and how to fix>
+
+Validation: <PASS|FAIL with output>
+```
+
+## Key Principles
+
+- **Binary verdict** - No "partial pass" or "mostly done"
+- **Evidence required** - Don't assume, verify
+- **Actionable feedback** - Every FAIL includes how to fix
+- **Criteria only** - Don't review quality, just correctness
+
+## Never Do
+
+- Pass with unmet criteria
+- Review code quality (that's quality-reviewer's job)
+- Make changes to the code
+- Add new requirements
+- Give subjective feedback

package/plugin/hooks/hooks.json

@@ -0,0 +1,3 @@
+{
+  "hooks": []
+}

package/plugin/references/review-criteria.md

@@ -0,0 +1,72 @@
+# Review Criteria Reference
+
+## Gap Severity Levels
+
+Use these severity levels when categorizing findings in review:
+
+| Severity | Meaning | Examples | Blocks Ship? |
+|----------|---------|----------|--------------|
+| CRITICAL | Security flaw, data loss, system crash | SQL injection, unencrypted secrets, null pointer crash | Yes |
+| HIGH | Missing feature, major bug, test failure | Required functionality missing, test suite fails | Yes |
+| MEDIUM | Missing edge case, incomplete docs | Edge case unhandled, missing error message | No |
+| LOW | Style, naming, minor polish | Inconsistent naming, verbose code | No |
+
+## Verdict Decision Matrix
+
+| Condition | Verdict |
+|-----------|---------|
+| Zero CRITICAL + Zero HIGH + All validation passes | SHIP |
+| Any HIGH (fixable) + No CRITICAL | FIX-FIRST |
+| Any CRITICAL or unfixable architectural issues | BLOCKED |
+
+## What Constitutes Evidence
+
+For PASS criteria, evidence includes:
+- Code exists that implements the feature
+- Test exists that verifies the behavior
+- Validation command succeeds with expected output
+- Documentation is present (when required)
+- Screenshot of working UI (for user-facing features)
+
+## Quality Review Categories
+
+### Security
+- Input validation present
+- Authentication/authorization correct
+- No injection vulnerabilities (SQL, XSS, command)
+- Secrets not hardcoded
+- OWASP Top 10 addressed
+
+### Maintainability
+- Code is readable
+- Appropriate abstraction level
+- Follows existing conventions
+- No dead code
+- No unresolved TODOs
+
+### Performance
+- No obvious N+1 queries
+- No unnecessary loops/allocations
+- Resources cleaned up
+- Appropriate data structures
+
+### Correctness
+- Edge cases handled
+- Error handling appropriate
+- Null/undefined safety
+- Type safety (if applicable)
+
+## Review Workflow Summary
+
+```
+/wish → creates plan
+/plan-review → validates plan structure
+/forge → executes plan
+/review → final validation → SHIP / FIX-FIRST / BLOCKED
+```
+
+## Fix Loop Limits
+
+- Spec review fix loops: max 3
+- Quality review fix loops: max 2
+- After max loops: mark task BLOCKED and continue

package/plugin/references/wish-template.md

@@ -0,0 +1,92 @@
+# Wish: <Title>
+
+**Status:** DRAFT | IN_PROGRESS | REVIEW | DONE
+**Slug:** `<slug>`
+**Created:** YYYY-MM-DD
+
+---
+
+## Summary
+
+<2-3 sentences: what we're building and why>
+
+---
+
+## Scope
+
+### IN
+- <what we're doing>
+- <specific deliverables>
+
+### OUT
+- <what we're explicitly not doing>
+- <boundaries to prevent scope creep>
+
+---
+
+## Decisions
+
+- **DEC-1:** <key decision with brief rationale>
+
+---
+
+## Success Criteria
+
+- [ ] <overall testable criterion 1>
+- [ ] <overall testable criterion 2>
+
+---
+
+## Assumptions
+
+- **ASM-1:** <assumption that if wrong, changes the plan>
+
+## Risks
+
+- **RISK-1:** <risk description> — Mitigation: <how we handle it>
+
+---
+
+## Execution Groups
+
+### Group A: <Name>
+
+**Goal:** <one sentence>
+
+**Deliverables:**
+- <concrete output 1>
+- <concrete output 2>
+
+**Acceptance Criteria:**
+- [ ] <testable criterion>
+- [ ] <testable criterion>
+
+**Validation:** `<command to verify completion>`
+
+---
+
+### Group B: <Name>
+
+**Goal:** <one sentence>
+
+**Deliverables:**
+- <concrete output>
+
+**Acceptance Criteria:**
+- [ ] <testable criterion>
+
+**Validation:** `<command to verify completion>`
+
+---
+
+## Review Results
+
+_Populated by `/review` after forge execution completes._
+
+---
+
+## Files to Create/Modify
+
+```
+<list of files this wish will touch>
+```