Loading...
Loading...
Found 23 Skills
Multi-source research synthesis — aggregate and compare 3+ sources or any source >5KB using sub-agent dispatch and SharedState
RED-GREEN-REFACTOR testing for agents: dispatch subagents with known inputs, capture verbatim outputs, verify against expectations. Use when creating, modifying, or validating agents and skills. Use for "test agent", "validate agent", "verify agent works", or pre-deployment checks. Do NOT use for feature requests, simple prompt edits without behavioral impact, or agents with no structured output to verify.
Track per-agent token usage and flag waste in parallel dispatch. Use after running parallel agents to evaluate cost vs value.
Dispatch background AI worker agents to execute tasks via checklist-based plans.
Plan a non-trivial code change end-to-end — size triage (XS–XL), slicing strategy, optional parallel subagent dispatch, per-slice Implement → Test → Verify → Commit discipline. Use for any multi-file change, refactor across files, executing a planned task from any planning source, cross-cutting modification (analytics sweep / i18n / library migration), or when about to write more than ~100 lines. 也用于增量实现 / 切片落地 / 推进已规划任务 / 跨切面改动。Skip only for trivial XS edits and pure documentation / configuration changes.
Run adversarial review on a PM artifact via the pm-critic sub-agent. Dispatches natively on Claude Code with the pm-skills plugin (invokes @agent-pm-critic); on non-Claude clients (Codex CLI, Cursor, Windsurf, Copilot, Gemini CLI) reads subagents/pm-critic.md and executes the system prompt inline. Returns findings graded P0/P1/P2/P3 with concrete fix suggestions per finding, plus a layered Status Summary section and machine-readable Status YAML block per master plan D26.
Concurrent investigation of independent failures. Use when multiple unrelated issues need parallel resolution.
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.
Mandatory orchestrator protocol - establishes ORCHESTRATOR principle (dispatch agents, don't operate directly) and skill discovery workflow for every conversation.
Concurrent investigation pattern - dispatches multiple AI agents to investigate and fix independent problems simultaneously.
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies