Loading...
Loading...
Found 334 Skills
Iterative worker-reviewer cycle that spawns a critic subagent to score work 1-10 and provide actionable feedback, then revises until a quality gate is met. Use when implementing features, writing specs, reviewing existing code, or completing any task where quality matters more than speed. Trigger phrases: "use review-loop", "polish this", "iterate on this", "/review-loop", "review with feedback loop".
Delegate noisy investigation to one or more subagents so the orchestrator's context stays clean, then work from the distilled answer. Use this skill whenever answering a question would require reading many files, long logs, large diffs, or wide codebase surveys — i.e. when producing the answer generates far more noise than the answer itself. Use it for "how does X work", "where is Y used", "what's the root cause of Z", "summarize this PR/log" style questions, and reach for it liberally before reading a pile of files inline.
Diagnose why an agent harness misbehaved by reading the local flight-recorder ledger (.vigiles/runs.jsonl) — which skills fired or got hijacked, which hooks blocked or wrongly allowed, which subagent tool-contract violations happened, and how a skill's trigger rate moved. Use when asked why a skill stopped firing, why a hook didn't block, why the wrong skill ran, or to debug/investigate what the harness actually did. NOT for writing new rules (use strengthen) or editing the spec (use edit-spec).
Validate whether an implementation matches its stated goal. Use this skill when a skill or agent wants a second opinion on its own output, when the user says "check this implementation", "validate what you did", "is this correct?", "review the output", or "did you do this right?". Also spawned automatically as a subagent by other skills (memory-bridge, daily-update) to self-check their outputs before presenting to the user. Returns a structured pass/warn/fail verdict with specific actionable issues.
Use when running Claude Fable on codebase-heavy or token-heavy work and the user wants Fable to orchestrate research, coding, and testing while cheaper subagents do bounded heavy lifting.
Inspect Claude Code session logs, tool calls, token usage, subagents, and context window using claude-devtools visual UI
Orchestrate multi-phase development workflows with strict role separation between implementers and validators. Automatically executes plans using separate subagents for implementation, validation, and fixing with auto-retry loops. Use when building complex systems requiring (1) Multi-step sequential or parallel development phases, (2) Automated validation with typecheck/build/tests after each phase, (3) Auto-retry fix loops until validation passes, (4) Complete execution after single user approval. Triggers include "implement this multi-phase plan", "build a system with phases", "create [complex system] following this architecture", "automate development workflow with validation", or any request for orchestrated development with multiple phases and quality checks. NOT for simple single-file tasks or exploratory coding.
Orchestrate Devin CLI subagents as background workers using tmux windows. Use when the user asks to spawn, coordinate, fan-out, or delegate work to multiple parallel agents, run background Devin sessions, or orchestrate long-running autonomous tasks from inside an existing Devin session.
Session mode: act as orchestrator brain only. Research and implementation go to cheaper-model subagents; the orchestrator scopes, briefs, verifies, and judges. User-invoked with the task as the argument.
Evaluate the reproducibility of technical articles. Dispatch a subagent to simulate a first-time reader reproducing the work locally and list missing information. Use as the final check on a draft before publication.
Carve guardrail-adjacent items out of scope with safe alternatives before risk-adjacent work starts, then run the safe remainder at full strength in a fresh subagent that only ever sees the carved prompt, never the risky input. Use when a task includes stealth, scraping, privacy, IP, policy, licensing, security, or other safety-adjacent material that could be silently dropped, over-elaborated, or needlessly diluted. Fires on the impulse, not only the topic: the moment you notice yourself about to hedge, soften, silently skip, or brace for a refusal, carve before you execute.
Spawn three parallel review subagents over the active transcript, surface learnings, and route each to a concrete edit on an existing skill. Use when the user says reflect.