Skip to content

prekuter/dryforge / agent-plugin/skills/go

Carry out the intent approved in `ready`, as meant, and prove it with checks that actually ran. Comes back to you if the work would change what you approved, keeps the project docs in step, and never merges on its own. Use when the user invokes the `go` skill after `ready`. Requires git.

references/reviewer-prompt.md

View on GitHub
# reviewer-prompt.md — final review (spec + code + harness)

After all waves merge, the completion gate passes, and the **harness has been created/updated**
(`harness-lifecycle.md`), one reviewer subagent checks the **full diff on the base** (from initial
state to current) **plus the harness**. This is the single review pass — spec conformance, code
quality, and (when the harness was created/updated this cycle) harness content and format.

> You are in a fresh session with no live user conversation — do **not** ask the user directly.
> Escalate via your structured return; the orchestrator relays escalations to the user.

## Scope — four lenses, one pass

**Lens 1: spec conformance.** Does the implementation do what the spec says — behavior, invariants,
edge-case rules, API surface? Every spec requirement should be traceable to code in the diff. Flag
missing behavior, violated invariants, edge cases the spec specifies that the code doesn't handle.

**Lens 2: code quality.** Cross-task consistency, seam leaks where tasks meet, duplication, naming
and pattern divergence across independently-written code. The completion gate already proved the
combined state builds and runs — your scope is what mechanical gates cannot see.

**Lenses 3–4: harness** (all four dimensions of `harness-review.md`, not only content/format) —
apply only when the harness was created
or updated this cycle. Do **not** inline harness criteria here; apply the four dimensions in
`references/harness-review.md` (provided with your dispatch) — content (substantive density + quality
principles), format (self-containment, altitude, no references), completeness (required files
present), and source-cross-check (omission vs. hallucination, future-scope content exempt). Using the
shared `harness-review.md` keeps a single source of truth — `migration` verifies against the same
criteria. Your dispatch states the user's language; flag a harness not written natively in it.
Harness findings carry the same blocking/advisory split as code findings.

## No fixed checklist — derive the rubric

Do **not** hardcode a quality checklist (that is a ceiling). Derive the rubric from the **spec**
(what matters for this feature) and the **project's conventions** (how the existing code is
written), then review against those.

## Calibration

Flag what would cause **real problems** — spec deviations, correctness, maintainability, convention
breaks that matter. Don't nitpick style the project doesn't care about. Separate **blocking** issues
(fix before proceeding) from **advisory** (note, non-blocking).

## Build-green blind spots

A green build (exit 0) does not guarantee the product works. Actively check for:
- **Declared assets exist on disk.** If the spec or config references files (icons, manifests,
  certificates, seed data), verify they exist in the build output — a missing asset is blocking.
- **Cross-boundary contract coverage.** If the project has both a server and a client (or multiple
  services), check whether automated tests verify the contract between them (route paths, request/
  response shapes). Pure-mock client tests + pure-unit server tests leave the integration seam
  untested. Flag the gap if no contract/integration test exists — advisory at minimum, blocking if
  the spec has explicit API invariants.

## Structured return

- `status`: `approved` | `issues`
- `issues`: blocking items, each with location and the fix
- `advisory`: non-blocking suggestions
rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 198kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 198kUpdated Claude CodeCodexVerifiedInstall