Каталог навыков агентов

This skill should be used when the user asks to "implement a DiD regression", "write a causal inference pipeline", "set...

Apply causal inference methods for treatment effect estimation and uplift modeling

This skill generates interactive Causal Loop Diagram (CLD) MicroSims using the vis-network JavaScript library. Use this...

Encode cause-effect logic into executable steps; use when asked to ensure robust troubleshooting and remediation.

Causal mediation analysis to identify which model components mediate specific behaviors. Use when investigating how info...

Causal reasoning in historical analysis. Covers distinguishing immediate from underlying causes, tracing chains of causa...

Systematic Fishbone analysis exploring problem causes across six categories

Use when auditing a CAV (Computer Aided Verification) submission for portal readiness, covering the four submission cate...

Use when planning a CAV (Computer Aided Verification) project timeline from venue and category selection through submiss...

Decision guide for delegating to caveman-style subagents. Tells the main thread WHEN to spawn `cavecrew-investigator` (l...

Use when delegating to caveman subagents (investigator/builder/reviewer) to shrink tool-result context.

When to delegate to `cavecrew-investigator` (locate code), `cavecrew-builder` (1-2 file edit) or `cavecrew-reviewer` (di...

Ultra-compressed communication mode. Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping...

Enables persistent ultra-compressed technical communication; use for explicit brevity or token-reduction requests, excep...

Globally persistent compressed response style with lite, full, ultra, and wenyan levels, plus off for normal prose. Trig...

Persistent terse-output overlay with safety and verification escape hatches. Use when the user explicitly says caveman o...

Ultra-compressed communication mode. Cuts output tokens 65% (measured) by speaking like caveman while keeping full techn...

Terse output mode — strips filler/articles/hedging, ~75% token reduction. Activate whenever: context is filling up, user...

Compressed communication mode. Drops filler, hedging, and narration while keeping full sentences and technical accuracy....

Persistent token-compression mode (~75% reduction) — now always-on by default. Strips filler while keeping code exact.

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy.

Mode de réponse compressé (inspiré de JuliusBrussee/caveman et GabrielBarberini/laconic) qui supprime le remplissage, le...

Ultra-compressed communication mode — cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping...

Ultra-compressed communication mode that preserves technical accuracy while dropping filler. Use when the user explicitl...

Cut ~75% of output tokens by talking like caveman. Full technical accuracy. Lite, full, ultra, and 文言文 intensity levels....

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accurac...

Ultra-compressed communication mode that cuts ~75% of token use by dropping articles, filler words, and pleasantries whi...

Caveman encoding for SPEC.md and spec-adjacent writes. Loaded by /spec, /build, /check. Cuts tokens ~75% vs prose while...

Ultra-compressed replies, no filler or hedging, technical accuracy kept. Use when the user says "caveman mode", "spartan...

Verbosity-reduction response register. Drops redundant clauses, narrative hedging, and ceremony while preserving article...

Use when the user says "caveman mode", "talk like caveman", "use caveman", "less tokens", or "be brief", invokes /cavema...

Ultra-compressed communication mode — drops filler, articles, pleasantries while keeping technical accuracy. Use when us...

Compression mode for ai-dev-team output. When active, the orchestrator drops articles, filler, hedging-as-decoration; pr...

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra...

Apply mandatory lite compression to internal AI communication without losing technical meaning or public clarity.

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like c...

User-invoked persistent ultra-compressed communication mode that targets 20-50% token reduction on typical replies while...

Terse caveman mode + companions. ONLY on explicit $caveman, /caveman, 'caveman mode', 'Caveman modunu aç'. Never for ord...

Respond terse like smart caveman. All technical substance stay. Only fluff die.

Terse, low-token responses. Minimal words, no fluff. Full capabilities preserved. Use when: optimize token usage, low-to...

Automatically activate token-saving caveman/cavecrew/caveman-compress patterns for any software engineering task. Use by...

Write a Conventional Commits message compressed to intent only. Use for "write a commit", "commit message", /commit or /...

Ultra-compressed commit message generator. Cuts noise from commit messages while preserving intent and reasoning. Conven...

Write a terse, exact Conventional Commits message from the intended diff. Use when asked for a commit message, when `/co...

Manual Caveman-style Conventional Commit message generator that preserves intent, rationale, breaking-change details, an...

Compress a memory file such as CLAUDE.md or a todo list into caveman format to save input tokens, keeping a readable bac...

Compress natural language memory files (CLAUDE.md, todos, preferences) into caveman format to save input tokens. Preserv...

Manual Caveman-style prose compressor for Markdown or plain-text memory and instruction files, with invariant validation...

Find and label every LLM workflow in the repository so Caveman Cloud groups spend by workflow instead of one bucket. Use...

Read-only review of Caveman Cloud evidence: cost, Cave Score, workflows, traces, latency, errors, routing, savings. Use...

Read-only repository explorer for cold-start orientation, broad cross-file localization, or when a direct search failed....

Quick-reference card for the three caveman skills and their commands. Trigger: /caveman-help or "caveman help".

Quick-reference card for caveman modes, skills and commands. Trigger: /caveman-help or "caveman help".

Quick-reference card for all caveman modes, skills, and commands. One-shot display, not a persistent mode. Trigger: /cav...

Manual one-shot reference card for Caveman levels and companion skills. Use only when explicitly invoked as $caveman-hel...

Act on a Caveman learn report - review the ranked token sinks, apply cost-lowering fixes with per-edit consent, and repo...

Inspect Caveman Cloud's experiment lifecycle and block unsafe execution. Use when asked to start, approve, cancel, promo...

Cut an ad, landing page, signup flow or any single screen down until a stranger gets it in 3 seconds: one picture, at mo...

Ultra-compressed communication mode. Cuts token usage ~75%. Use when user requests caveman mode, less tokens, brief talk...

Turn a Caveman optimization observation into an operator-chosen candidate with a paired baseline evaluation. Use when as...

Ultra-compressed code review comments. Cuts noise from PR feedback while preserving the actionable signal. Each comment...

Compressed code review - one line per finding with location, problem and fix. Use for /caveman-review, "review this PR",...

Manual Caveman-style code-review mode that produces concise findings with exact locations, evidence, impact, and a concr...

Wire a repository through the Caveman Cloud gateway so every LLM request is measured, with no behavior change. Use for "...

Show recorded output and cache-read token usage and mode attribution for the current Claude Code session, or locate the...

Show real token usage and estimated savings for the current session, read from the session log. Trigger: /caveman-stats.

Show real token usage and estimated savings for the current session. Reads directly from the Claude Code session log — n...

Compress technical plans, findings and handoffs into dense AI-oriented shorthand when explicitly requested; preserve con...

可转债全量列表（market.ft.tech）。用户问可转债列表、全部可转债、转债代码列表、有哪些可转债时使用。

College Basketball (CBB) data via ESPN public endpoints — scores, standings, rosters, schedules, game summaries, play-by...

Query the China Biographical Database (CBDB) API to retrieve comprehensive biographical data about historical Chinese fi...

Read-only Central Bank of Iran (cbi.ir): official FX rates, inflation, policy rates, news. Use when the user asks about...

Query cBioPortal for cancer genomics data including somatic mutations, copy number alterations, gene expression, and sur...

Query the Yana AI codebase knowledge graph via codebase-memory-mcp. Use instead of grep/glob when exploring call chains,...

Comprehensive guide to cbow. Master the concepts, implementation, best practices, and real-world applications of cbow in...

Plan CBRN response and consequence management actions. Use when preparing detection response drills, contamination contr...

Make a YouTube video accessible: WCAG 2.1 / Section 508 corrected captions, audio description and accessible videos.

Short slash command wrapper for Claude relay sessions. Provides /cc commands like /cc on <project>, /cc tail, /cc off, a...

Short slash command wrapper for Claude relay sessions. Use when the user wants concise /cc commands like /cc projects, /...

Claude × Codex × GPT-5.5 Pro × Gemini 3 Pro 多模型协作流水线(Anthropic/OpenAI/Google 三族交叉互审)。当 Tony 说「cc」「走 cc」「用 cc 做」「cc 这个任务」...

Print a ready-made /compact line in a canonical handoff format, so the operator pastes it as the next message and shrink...

Claude Code 專案配置審計。觸發：review/優化 CLAUDE.md、skills、settings、定期清洗累積內容、新專案上線前檢查。

Agent 配置設計指南 — 基於 Claude Code 6 個 built-in agent 的逆向分析。Use when: 設計新 agent、優化現有 agent prompt、決定工具/模型配置、撰寫 dispatch promp...

Delegates coding/research tasks to the Google Antigravity CLI (`agy`) for external-model execution (Gemini 3.x, Claude S...

Use when user asks for Claude Code usage stats, weekly analytics, project activity summary, or wants to see what project...

当设计或修改 REST API 响应结构、处理 API 返回值，或生成 Excel/CSV/PDF/对账文件等下游产物时触发。防止 API 设计缺陷导致的字段错位、类型歧义，以及生成产物时关键字段缺失但静默成功的问题。

网关/代理/WAF/CDN 中间件响应判定与关键词匹配安全规范。适用于 proxy/gateway/waf/cdn/nginx/openresty/rewrite/redirect 等代码,防止把正常长正文中的 Cloudflare、502...

Use when designing AskUserQuestion tool calls - covers schema constraints, effectiveness patterns, and known issues.

Bash、Shell、Dockerfile、Makefile 与命令片段规范，适用于脚本修改、命令编写和文档中的 shell 示例；不负责运维审批策略。

Use when Pure C module interfaces, ownership contracts, error-code cleanup paths, build conventions or C engineering val...

CONTRIBUTOR TOOL - Track CC changelog, extract new versions since last check, analyze impact on plugin (breaking changes...

Review code formatting and documentation using 74-item checklist from Code Complete. Two modes: CHECKER executes checkli...

Delegates coding tasks to Codex CLI for prototyping, debugging, and code review. Use when: (1) Backend/logic implementat...

CC-Codex 协作讨论。自由话题驱动的 CC-Codex 协作工具，Battle Loop 辩论机制。关键词: 审查, review, 让codex看看, codex审查, 帮我审查, 讨论, discuss, battle, 让cod...

结构化 git commit message 生成工作流，适用于显式 commit message、提交说明或 commit-msg 场景；不负责实际执行 git commit。

Use when adding rules, commands, or skills to dotfiles - decides global vs project-local placement