skills/review/SKILL.md
Third-party review of finished Claude sessions. Reads completion signals written by the plugin's Stop hook, analyzes each session's transcript for patterns (skill-heavy/code-light loops, duplicate skill invocations, context compaction thrash), writes verdicts to .harness/metrics/, and appends learnings to .harness/learnings/knowledge.md. Use this when you want to audit what happened in recent sessions and feed learnings back into chain config.
npx skillsauth add Cheggin/skill-chain reviewInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Runs the bundled reviewer against all pending completion signals, or a specific signal file.
The Stop hook (hooks/completion-signal.mjs) writes .harness/signals/done-<sessionId>.json whenever a Claude session ends. Those signals sit until a reviewer processes them.
The reviewer is intentionally out-of-process from the session that produced the transcript — self-review is biased. It reads the transcript JSONL from the signal payload, computes metrics, and emits a verdict (clean / warn / fail).
Review every pending signal:
node ${CLAUDE_PLUGIN_ROOT}/scripts/review.mjs --all-pending
Review a specific signal:
node ${CLAUDE_PLUGIN_ROOT}/scripts/review.mjs .harness/signals/done-<sessionId>.json
After review:
.harness/metrics/<sessionId>.json.harness/learnings/knowledge.md.harness/signals/processed/ (won't be re-reviewed)After a long autonomous run, a dispatched tmux agent, or at end of day when you want to look at what patterns emerged. Findings are informational by default; risky patterns (duplicate skill calls, plan-only runs with no code output, repeated compactions) are flagged explicitly so the next chain iteration can address them.
development
Design engineering principles for making interfaces feel polished. Use when building UI components, reviewing frontend code, implementing animations, hover states, shadows, borders, typography, micro-interactions, enter/exit animations, or any visual detail work. Triggers on UI polish, design details, "make it feel better", "feels off", stagger animations, border radius, optical alignment, font smoothing, tabular numbers, image outlines, box shadows.
documentation
Agentic memory system for writers - track characters, relationships, scenes, and themes
documentation
LLM Wiki — persistent markdown knowledge base that compounds across sessions (Karpathy model)
development
Build production-quality SaaS websites with opinionated design presets. Use when creating any startup website. The user MUST pick a design style before building. Enforces shadcn/ui, Figma design principles, specific CSS values per style, and anti-AI-writing. Load alongside anti-ai-writing skill. LIGHT MODE ONLY.