Announce: "I'm using hyperpowers:sre-task-refinement to review this plan with Google Fellow-level scrutiny."

Review Checklist (Apply to Every Task)

1. Task Granularity

Check:

• No task >8 hours (subtasks) or >16 hours (phases)?
• Large phases broken into 4-8 hour subtasks?
• Each subtask independently completable?
• Each subtask has clear deliverable?

If task >16 hours:

• Create subtasks with bd create
• Link with bd dep add child parent --type parent-child
• Update parent to coordinator role

2. Implementability (Junior Engineer Test)

Check:

• Can junior engineer implement without asking questions?
• Function signatures/behaviors described, not just "implement X"?
• Test scenarios described (what they verify, not just names)?
• "Done" clearly defined with verifiable criteria?
• All file paths specified or marked "TBD: new file"?

Red flags:

• "Implement properly" (how?)
• "Add support" (for what exactly?)
• "Make it work" (what does working mean?)
• File paths missing or ambiguous

3. Success Criteria Quality

Check:

• Each task has 3+ specific, measurable success criteria?
• All criteria testable/verifiable (not subjective)?
• Includes automated verification (tests pass, clippy clean)?
• No vague criteria like "works well" or "is implemented"?

Good criteria examples:

• ✅ "5+ unit tests pass (valid VIN, invalid checksum, various formats)"
• ✅ "Clippy clean with no warnings"
• ✅ "Performance: <100ms for 1000 records"

Bad criteria examples:

• ❌ "Code is good quality"
• ❌ "Works correctly"
• ❌ "Is implemented"

4. Dependency Structure

Check:

• Parent-child relationships correct (epic → phases → subtasks)?
• Blocking dependencies correct (earlier work blocks later)?
• No circular dependencies?
• Dependency graph makes logical sense?

Verify with:

bd dep tree bd-1  # Show full dependency tree

5. Safety & Quality Standards

Check:

• Anti-patterns include unwrap/expect prohibition?
• Anti-patterns include TODO prohibition (or must have issue #)?
• Anti-patterns include stub implementation prohibition?
• Error handling requirements specified (use Result, avoid panic)?
• Test requirements specific (test names, scenarios listed)?

Minimum anti-patterns:

• ❌ No unwrap/expect in production code
• ❌ No TODOs without issue numbers
• ❌ No stub implementations (unimplemented!, todo!)
• ❌ No regex without catastrophic backtracking check

6. Edge Cases & Failure Modes (Fellow SRE Perspective)

Ask for each task:

• What happens with malformed input?
• What happens with empty/nil/zero values?
• What happens under high load/concurrency?
• What happens when dependencies fail?
• What happens with Unicode, special characters, large inputs?
• Are these edge cases addressed in the plan?

Add to Key Considerations section:

• Edge case descriptions
• Mitigation strategies
• References to similar code handling these cases

7. Red Flags (AUTO-REJECT)

Check for these - if found, REJECT plan:

• ❌ Any task >16 hours without subtask breakdown
• ❌ Vague language: "implement properly", "add support", "make it work"
• ❌ Success criteria that can't be verified: "code is good", "works well"
• ❌ Missing test specifications
• ❌ "We'll handle this later" or "TODO" in the plan itself
• ❌ No anti-patterns section
• ❌ Implementation checklist with fewer than 3 items per task
• ❌ No effort estimates
• ❌ Missing error handling considerations
• ❌ CRITICAL: Placeholder text in design field - "[detailed above]", "[as specified]", "[complete steps here]"

Review Process

For each task in the plan:

Step 1: Read the task

bd show bd-3

Step 2: Apply all 7 checklist categories

• Task Granularity
• Implementability
• Success Criteria Quality
• Dependency Structure
• Safety & Quality Standards
• Edge Cases & Failure Modes
• Red Flags

Step 3: Document findings Take notes:

• What's done well
• What's missing
• What's vague or ambiguous
• Hidden failure modes not addressed
• Better approaches or simplifications

Step 4: Update the task

Use bd update to add missing information:

bd update bd-3 --design "$(cat <<'EOF'
## Goal
[Original goal, preserved]

## Effort Estimate
[Updated estimate if needed]

## Success Criteria
- [ ] Existing criteria
- [ ] NEW: Added missing measurable criteria

## Implementation Checklist
[Complete checklist with file paths]

## Key Considerations (ADDED BY SRE REVIEW)

**Edge Case: Empty Input**
- What happens when input is empty string?
- MUST validate input length before processing

**Edge Case: Unicode Handling**
- What if string contains RTL or surrogate pairs?
- Use proper Unicode-aware string methods

**Performance Concern: Regex Backtracking**
- Pattern `.*[a-z]+.*` has catastrophic backtracking risk
- MUST test with pathological inputs (e.g., 10000 'a's)
- Use possessive quantifiers or bounded repetition

**Reference Implementation**
- Study src/similar/module.rs for pattern to follow

## Anti-patterns
[Original anti-patterns]
- ❌ NEW: Specific anti-pattern for this task's risks
EOF
)"

IMPORTANT: Use --design for full detailed description, NOT --description (title only).

Step 5: Verify no placeholder text (MANDATORY)

After updating, read back with bd show bd-N and verify:

• ✅ All sections contain actual content, not meta-references
• ✅ No placeholder text like "[detailed above]", "[as specified]", "[will be added]"
• ✅ Implementation steps fully written with actual code examples
• ✅ Success criteria explicit, not referencing "criteria above"
• ❌ If ANY placeholder text found: REJECT and rewrite with actual content

Breaking Down Large Tasks

If task >16 hours, create subtasks:

# Create first subtask
bd create "Subtask 1: [Specific Component]" \
  --type task \
  --priority 1 \
  --design "[Complete subtask design with all 7 categories addressed]"
# Returns bd-10

# Create second subtask
bd create "Subtask 2: [Another Component]" \
  --type task \
  --priority 1 \
  --design "[Complete subtask design]"
# Returns bd-11

# Link subtasks to parent with parent-child relationship
bd dep add bd-10 bd-3 --type parent-child  # bd-10 is child of bd-3
bd dep add bd-11 bd-3 --type parent-child  # bd-11 is child of bd-3

# Add sequential dependencies if needed (LATER depends on EARLIER)
bd dep add bd-11 bd-10  # bd-11 depends on bd-10 (do bd-10 first)

# Update parent to coordinator
bd update bd-3 --design "$(cat <<'EOF'
## Goal
Coordinate implementation of [feature]. Broken into N subtasks.

## Success Criteria
- [ ] All N child subtasks closed
- [ ] Integration tests pass
- [ ] [High-level verification criteria]
EOF
)"

Output Format

After reviewing all tasks:

## Plan Review Results

### Epic: [Name] ([epic-id])

### Overall Assessment
[APPROVE ✅ / NEEDS REVISION ⚠️ / REJECT ❌]

### Dependency Structure Review
[Output of `bd dep tree [epic-id]`]

**Structure Quality**: [✅ Correct / ❌ Issues found]
- [Comments on parent-child relationships]
- [Comments on blocking dependencies]
- [Comments on granularity]

### Task-by-Task Review

#### [Task Name] (bd-N)
**Type**: [epic/feature/task]
**Status**: [✅ Ready / ⚠️ Needs Minor Improvements / ❌ Needs Major Revision]
**Estimated Effort**: [X hours] ([✅ Good / ❌ Too large - needs breakdown])

**Strengths**:
- [What's done well]

**Critical Issues** (must fix):
- [Blocking problems]

**Improvements Needed**:
- [What to add/clarify]

**Edge Cases Missing**:
- [Failure modes not addressed]

**Changes Made**:
- [Specific improvements added via `bd update`]

---

[Repeat for each task/phase/subtask]

### Summary of Changes

**Issues Updated**:
- bd-3 - Added edge case handling for Unicode, regex backtracking risks
- bd-5 - Broke into 3 subtasks (was 40 hours, now 3x8 hours)
- bd-7 - Strengthened success criteria (added test names, verification commands)

### Critical Gaps Across Plan
1. [Pattern of missing items across multiple tasks]
2. [Systemic issues in the plan]

### Recommendations

[If APPROVE]:
✅ Plan is solid and ready for implementation.
- All tasks are junior-engineer implementable
- Dependency structure is correct
- Edge cases and failure modes addressed

[If NEEDS REVISION]:
⚠️ Plan needs improvements before implementation:
- [List major items that need addressing]
- After changes, re-run hyperpowers:sre-task-refinement

[If REJECT]:
❌ Plan has fundamental issues and needs redesign:
- [Critical problems]