prd-implementation-precheck — Claude Code skill: PRD check before code
This skill runs a PRD or feature spec through a checklist before implementation, surfacing edge-case gaps and blockers, then gets on with the build.
npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a claude-codeWhat prd-implementation-precheck does
The skill locates the PRD and any files it references, sums up the intent in a sentence or two, and produces a report: what is covered, what is missing, where the blockers are. From there it either asks questions or states its assumptions and implements. It only asks when a missing choice would change behavior, architecture, data, or external side effects.
The checklist has layers. Basic checks cover scope, conflicts with existing patterns, dependencies, behavior, and risk. Separate blocks handle edge cases: null and empty inputs, boundary values, duplicates, state transitions, rollbacks, network failures, timeouts, partial failures, empty and loading states, permission denied. One more block scans the codebase to see whether the PRD matches the delete strategy, error handling, component reuse, and API conventions already in use.
The report follows a fixed template: summary, covered edge cases, a table of missing ones with a suggested default and a confirmation flag, blockers, warnings, questions. The closing prompt about proceeding as-is or updating the PRD only appears when the answer changes the implementation. No blockers means the skill states assumptions and moves on without a redundant confirmation round. It needs file read and edit tools plus Bash, Grep, and Glob — without them the codebase consistency pass cannot run.
What it helps with
- find gaps in a PRD before any code is written
- separate hard blockers from optional warnings
- check requirements against existing code patterns
- verify edge cases and rollback paths are specified
- collect questions for the spec author
Example prompts for the agent
- precheck the PRD in docs/prd-checkout.md before implementing
- run a precheck on this feature spec and list the blockers
- check this PRD for missing edge cases and continue if there are no blockers
How to install prd-implementation-precheck
1Skills CLI (Vercel)
npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a claude-code2Manually: put the files in place
~/.claude/skills/prd-implementation-precheck/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/prd-implementation-precheck/
1Skills CLI (Vercel)
npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a codex2Manually: put the files in place
.agents/skills/prd-implementation-precheck/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a cursor2Manually: put the files in place
.cursor/skills/prd-implementation-precheck/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install zhaono1/agent-playbook prd-implementation-precheck2Manually: put the files in place
.github/skills/prd-implementation-precheck/SKILL.md1Skills CLI (Vercel)
npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a gemini-cli2Platform CLI
gemini skills install https://github.com/zhaono1/agent-playbook --path skills/prd-implementation-precheckManually: put the files in place
.devin/skills/prd-implementation-precheck/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about prd-implementation-precheck
How do I install prd-implementation-precheck in Claude?
Run npx skills add zhaono1/agent-playbook --skill prd-implementation-precheck -a claude-code. The skill lands at ~/.claude/skills/prd-implementation-precheck/SKILL.md.
What does prd-implementation-precheck do?
It checks a PRD or feature spec for scope, dependencies, behavior, risks, edge cases, and alignment with existing code patterns. Then it either asks about blockers or states assumptions and starts implementing.
When does the skill ask the user a question?
Only when a missing choice would materially change behavior, architecture, data, or external side effects. With no blockers it states assumptions and continues without a confirmation round.
Source text
SKILL.md · MIT
---
name: prd-implementation-precheck
description: Precheck a PRD or feature specification before implementation, resolving material blockers while continuing directly when requirements are clear.
allowed-tools: Read, Write, Edit, Bash, Grep, Glob, AskUserQuestion
metadata:
hooks:
after_complete:
- trigger: self-improving-agent
mode: background
reason: "Learn from implementation patterns"
- trigger: session-logger
mode: auto
reason: "Log PRD implementation activity"
---
# PRD Implementation Precheck
## Overview
Perform a short PRD precheck, surface material issues, then implement. Ask the
user only when a missing choice would materially change behavior, architecture,
data, or external side effects.
## Workflow
1. Locate the PRD and any referenced files.
2. Precheck the PRD and summarize intent in 1-2 sentences.
3. List findings and questions with blockers first. If there are no material blockers, state assumptions and continue.
4. When a blocker exists, resolve it with the user before implementation.
5. Validate (tests or manual steps) or state what was not run.
## Precheck Checklist
### Basic Checks
- **Scope**: Identify over-broad changes; suggest a smaller, targeted approach.
- **Alignment**: Flag conflicts with existing patterns or architecture; propose alternatives.
- **Dependencies**: Note missing hooks/providers/data sources or unclear ownership.
- **Behavior**: Verify flows and edge cases are specified; ask for gaps.
- **Risks**: Call out performance, regressions, or migration risks.
- **Testing**: Check success criteria and test coverage; request specifics if vague.
### Edge Case Coverage Checks
Verify the PRD addresses these edge cases (mark as ⚠️ if missing):
#### Data Boundaries
- [ ] **Null/Empty handling** - What happens with empty inputs or null values?
- [ ] **Boundary values** - Are min/max limits defined? What happens at boundaries?
- [ ] **Duplicate data** - How are duplicates detected and handled?
- [ ] **Data format** - Are input formats validated? What about special characters?
#### State Boundaries
- [ ] **State transitions** - Are all valid state transitions defined?
- [ ] **Invalid transitions** - What happens on illegal state changes?
- [ ] **Concurrent modifications** - How are simultaneous edits handled?
- [ ] **Rollback scenarios** - Can operations be undone? How?
#### Error Boundaries
- [ ] **Network failures** - What happens when API calls fail?
- [ ] **Timeout behavior** - Are timeouts defined? What's the retry strategy?
- [ ] **Partial failures** - If step 2 of 3 fails, what happens to step 1?
- [ ] **Error messages** - Are user-facing error messages defined?
#### UX Boundaries
- [ ] **Empty states** - What does the user see with no data?
- [ ] **Loading states** - How is loading indicated?
- [ ] **Success feedback** - How does the user know the action succeeded?
- [ ] **Permission denied** - What happens when user lacks permission?
### Codebase Consistency Checks
Scan the codebase to verify PRD aligns with existing patterns:
```bash
# Check if PRD's proposed patterns match existing code
grep -r "pattern_from_prd" src/ --include="*.ts"
```
- [ ] **Delete strategy** - Does PRD match existing soft/hard delete pattern?
- [ ] **Error handling** - Does PRD use the same error display mechanism?
- [ ] **Component reuse** - Does PRD leverage existing components?
- [ ] **API patterns** - Does PRD follow existing API conventions?
## Output Format
### Precheck Report Template
```markdown
## PRD Precheck Report
### Summary
{1-2 sentence summary of what the PRD aims to achieve}
### ✅ Covered Edge Cases
- {List edge cases that are well-defined in the PRD}
### ⚠️ Missing Edge Cases
| Edge Case | Category | Suggested Default | Needs Confirmation |
|-----------|----------|-------------------|-------------------|
| Empty list display | UX | Use existing EmptyState | No |
| Concurrent edit | State | Last write wins | **Yes** |
### 🔴 Blockers
- {Critical issues that must be resolved before implementation}
### 🟡 Warnings
- {Non-critical issues that should be addressed}
### Questions for User
1. {Specific question about missing edge case}
2. {Specific question about ambiguous requirement}
---
**Proceed as-is, or update the PRD?**
```
## Output Expectations
- Provide a concise precheck report with questions and risks.
- Ask "Proceed as-is, or update the PRD?" only when the answer changes the implementation materially.
- If there are no blockers, state assumptions and continue without a redundant confirmation round.
© zhaono1 · MIT · zhaono1/agent-playbook