Loading...
Loading...
Found 26 Skills
Превращает идею в конкретную задачу, которую агент может выполнить. Проводит через 5 шагов: идея → результат → реальность → задача → план.
Based on the Recursive Language Models (RLM) research by Zhang, Kraska, and Khattab (2025), this skill provides strategies for handling tasks that exceed comfortable context limits through programmatic decomposition and recursive self-invocation. Triggers on phrases like "analyze all files", "process this large document", "aggregate information from", "search across the codebase", or tasks involving 10+ files or 50k+ tokens.
Parallel Agent Orchestration
Hand off a task to GitHub Copilot.
Maintain a structured ledger of decisions, discovered bugs and fixes, user preferences, constraints, current status, and failed approaches throughout multi-step agentic tasks. Auto-update after every significant step. Triggers on "where were we", "continue", "summarize status", "remember", or when a new agent instance takes over a task.
MindOS Knowledge Base Operation Guide (Chinese) for Agent tasks on local markdown/csv knowledge bases. It should be automatically triggered whenever tasks involve note files, SOP/workflow documents, profile/context documents, CSV tables, knowledge base organization, cross-Agent handover or decision synchronization, and are executed via the MindOS MCP tool. Typical requests include "update notes", "search knowledge base", "organize files", "execute SOP", "review according to team standards", "hand over tasks to another Agent", "synchronize decisions", "append to CSV", "retrospect this conversation", "extract key experiences", "adaptively update retrospective results to corresponding documents", "route this information to corresponding files", "synchronously update all related documents", etc.; it should be triggered even if the user does not explicitly mention MindOS.
Break a design document into wave-ordered implementation tasks with domain agent assignments. Use after /feature-design produces a design doc. Use for "plan feature", "break down design", "create tasks", or "/feature-plan". Do NOT use without a design doc or for simple single-task work.
Use when creating or maintaining OpenCLI site sitemaps: agent-facing navigation, page-state, action, workflow, API-reference, pitfall, and fallback knowledge for a website. Use after browser exploration discovers durable site context, when a sitemap is stale, or when promoting local site knowledge into the repo.
Review applications and verify task submissions on OpenAnt. Use when the agent (as task creator) needs to review applicants, accept or reject applications, approve or reject submitted work, or give feedback on deliverables. Covers "review applications", "approve submission", "reject work", "check applicants", "verify task".
Create, query, update, assign, and discuss Multica issues. Also covers comments, subscribers, and viewing execution runs for an issue. Use when the user wants to file a task for an agent, triage the board, comment on an issue, or inspect what an agent actually did.
Inline risk classification for agent tasks using a 4-tier model. Hybrid routing: GREEN/YELLOW use heuristic file-pattern matching, RED/CRITICAL escalate to war-room-checkpoint for full reversibility scoring.
Use when an agent needs to write user stories for a project