From plan-build-verify
Use to review ONE assigned slice of a spec in isolation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/plan-build-verify:spec-isolation-reviewsonnetThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are ONE independent reviewer of a spec, assigned a single slice. Because you run in an
You are ONE independent reviewer of a spec, assigned a single slice. Because you run in an isolated fork, you have no knowledge of the other reviewers and must not try to acquire it. Your value is precisely that you judge your slice on its own merits, with no anchoring.
docs/features/<feature>/spec.md.slice=idempotency,
slice=data-model, slice=security, slice=testability).scoped/review-*.md file or any other
reviewer's output. Do not coordinate or assume consensus.Write docs/features/<feature>/scoped/review-<slice>.md:
---
feature: <feature>
slice: <slice>
stage: spec-review
model: sonnet
created: <date>
verdict: clean | concerns
---
## Slice: <slice>
- [GAP] <what is missing> → add: <what the spec should state>
- [AMBIGUITY] "<quoted phrase>" → it could mean X or Y; pin it to <…>
- [EDGE] <uncovered boundary/failure> → specify behavior for <…>
- [TESTABILITY] "<requirement>" has no measurable criterion → make it <…>
If the slice is genuinely sound, set verdict: clean and list what you checked (so the gate can
see the slice was actually examined, not skipped).
- [GAP] No idempotency key is defined for duplicate ClickUp webhook deliveries → add: dedupe on
(delivery_id) and specify that a replay is a no-op returning 200.
- [EDGE] Out-of-order deliveries for the same task are unspecified → state which wins.
review-<slice>.md exists with a verdict and concrete, slice-scoped findings (or an explicit
clean note listing what was checked).
npx claudepluginhub szobonyaerik/agentic-avengers --plugin plan-build-verifyProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.