evgenygurin

Use when batch-delegating multiple independent tasks to Codegen cloud agents in a single call. Uses codegen_bulk_create_runs for parallel execution. Best for plans where tasks have no inter-dependencies.

Use when debugging failed, stuck, or misbehaving Codegen agent runs — systematic 4-phase approach. Triggers on agent run failures, timeouts, or unexpected results. Analogous to superpowers systematic-debugging but for cloud agents.

Use when reviewing code and PRs created by Codegen agents — two-stage review process. Stage 1 checks spec compliance (did the agent do what was asked?). Stage 2 checks code quality (is the code good?). Triggers after agent runs complete.

Use when analyzing Codegen agent run performance — interpret success rates, duration metrics, token usage, and trends via codegen_get_run_analytics. Best used after batch runs or for periodic performance review.

Guide for writing effective prompts for Codegen cloud agents. Includes templates, anti-patterns, and optimization strategies. Auto-triggers when creating agent runs to ensure prompt quality.

Monitor running Codegen agents — poll status, review logs, detect blockers, and surface progress. Triggered when codegen_get_run, codegen_get_logs, codegen_list_runs, or codegen_get_execution_context tools are invoked.

Automatically delegate tasks to Codegen cloud agents when user requests code changes, implementations, or fixes that should run remotely. Triggered when codegen_create_run or codegen_start_execution tools are invoked. Includes prompt templates and quality checklist.

Use when executing implementation plans via Codegen cloud agents instead of locally. Delegates each task as a separate Codegen agent run, monitors progress, and handles blockers. Drop-in replacement for superpowers:executing-plans — works with the same plan format.

Use when managing DJ techno music library, building DJ sets, analyzing audio/transitions, or working with Yandex Music/Beatport/Suno providers. Triggers on keywords: dj, set, mix, track, bpm, camelot, transition, playlist, library, techno, suno, yandex music, beatport. Use ONLY when the task involves DJ-specific domain (library management, set building, audio analysis, provider sync). Do NOT use for general programming.

This skill should be used when the user asks to build a DJ set, create a set from playlist or library, optimize track order, rebuild set, reorder tracks, or make a set. Covers candidate curation, GA/greedy optimization, L5 finalization, review and iteration.

This skill should be used when the user asks to classify tracks, audit playlist, get library stats, distribute to subgenres, run mood classification, curate a crate, or review library quality. Covers feature-first curation, mood classification, audits, dedup and stats.

This skill should be used when the user asks to deliver a set, export a set, finalize a set, do a rekordbox export, sync a set to YM, or generate a cheat sheet. Covers the generated-sets bundle (MP3 + M3U8 + cheatsheet + rekordbox XML), YM playlist push and backup.

This skill should be used when the user asks to expand a playlist, find similar tracks, add more tracks, discover new tracks, crate digging, import from Yandex Music, or fill gaps in a playlist. Covers seed-based discovery, import, download and analysis.

This skill should be used when the user asks to sync a playlist, push or pull from Yandex Music, search YM, manage YM playlists, or manage YM likes. Covers bidirectional sync, playlist management, ordered pushes, search and likes.

Use when running an active autopilot-codex session — explains the plan→implement→review→verdict cycle, state machine, and termination conditions. Required reading when status='running' and the implementer needs to act inside the loop.

Use when interpreting state mutations during plan/review/verdict transitions — explains exactly which fields the bridge writes when, so the implementer can debug 'why is current_task suddenly T3' or 'why is base_sha old'.

Use when deciding whether a question warrants spawning a Codex turn or you can answer locally — covers the cost/value matrix and which subagent or command is the right entry point for each kind of consultation.

Use whenever you're about to run a destructive Bash command, push code, or spend Codex tokens — explains every cap and denylist this plugin enforces and how to extend them.

Use whenever the user invokes /autopilot-codex:* in this repo, OR mentions "autopilot", "Codex manager", "Codex architect", or asks Codex to plan/review for them. Establishes the role split — OpenAI Codex is manager/architect, Claude Code is implementer — and the operating contract.

Use when the user asks a crypto-research question that needs multiple tool calls (ticker + OHLCV, comparison, trend analysis, liquidity check). This skill explains the research loop, when to call which tool, and how to structure findings.

Use when the user brings a declarative trading strategy (StrategySpec payload, a saved strategy id, or a plain-language "review my strategy" request). This skill runs the validate → explain → backtest → compare/stress → save loop against the Phase-2 MCP tools and packages the findings in a Claude-friendly order.

Use at the end of a cryptozavr research session (or when the user asks "wrap up" / "what did we do"). Produces a disciplined 3-bullet summary — artifacts produced, decisions made, what's next — so the next session starts warm.

Use when a cryptozavr tool returns degraded/stale data, times out, or reports an upstream failure (KuCoin/CoinGecko). Walks through the L2 infrastructure chain, inspects provider metrics, and pinpoints whether the issue is rate limiting, transport, provider downtime, or cache staleness.

Use when you've just received output from a cryptozavr MCP tool (ticker, OHLCV, orderbook, trades) and need to read it correctly. Covers field meanings, staleness flags, reason_codes, and common pitfalls.

Use when building DJ sets, scoring transitions, delivering sets, syncing to YM playlists, or improving set quality. Triggers on build_set, deliver_set, score_transitions, rebuild_set, "построй сет", "экспортируй сет".

Use when working with audio features, BPM, Camelot keys, transition scoring, GA set builder, iCloud stubs, M3U8 export, or cheat_sheet generation. Triggers on app/utils/audio/, TrackFeatures, scoring, deliver_set, cheat_sheet.

Use when delegating tasks to Codegen cloud agents, monitoring agent runs, reviewing agent PRs, or running multiple agents in parallel. Triggers on "delegate", "codegen", /delegate, "@codegen-sh", Codegen Bridge.

Use when adding a new MCP tool, registering tools in gateway, creating Pydantic types for structured_content, or writing MCP tool tests. Triggers on app/mcp/tools/, FastMCP, register_tools, "new tool", "add instrument".

Use when expanding playlists with new tracks, discovering similar tracks, importing from YM, auditing quality, or distributing by subgenre. Triggers on "расширь плейлист", "найди похожие", "заполни плейлист", "expand playlist", "fill playlist", discover_candidates, audit_playlist, distribute_to_subgenres.

Automatically search and retrieve relevant project memory from Mem0 when working on tasks that might benefit from past decisions, patterns, or learnings. Use this skill when starting new features, fixing bugs, or making architectural decisions to leverage historical context.

Use when collaborating with Codex CLI (via codex-bridge) on a non-trivial feature, refactor, or investigation — typically triggered by /codex-collab, "работай с codex", "делегируй исследование codex", "codex + ты над X", or any phrasing that requests shared work between Claude Code and Codex. Enforces a research → plan → implement → verify → fix/docs → commit workflow with session_id continuity so Codex serves as analyst/reviewer while Claude Code remains the sole code editor. Delegates multi-file exploration to Codex to keep Claude Code's context window narrow.

This skill enriches vague prompts with targeted research and clarification before execution. Should be used when a prompt is determined to be vague and requires systematic research, question generation, and execution guidance.