菜单
Compare autoresearch results across runs, branches, or machines. Reads results.tsv from multiple branches and generates comparison tables. Use when the user asks "compare runs", "how did machine X vs Y", or "cross-run analysis".
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Master orchestrator for 3D/2D game development — engine selection, architecture, asset pipelines, physics, multiplayer, and deployment. Routes to 10 Three.js sub-skills and 7 reference guides.
ANIMA Manifest Designer — Designs, validates, and evolves the three core YAML standards that define the ANIMA ecosystem: anima_module.yaml, anima_stack.yaml, anima_program.yaml. This is the "npm package.json of robotics" — whoever defines this standard owns the ecosystem. Modes: MODULE (design module manifest), STACK (design stack manifest), PROGRAM (design program manifest), VALIDATE (check a manifest against the schema), PROFILE (define data profiles).
Mass marketing automation for RobotFlow Labs projects. 8 modes: SUBMIT (awesome lists, GitHub Discussions), SOCIAL (X/Twitter, LinkedIn, Reddit drafts), HN (Hacker News Show HN), REDDIT (browser-post to subreddits), BLOG (Dev.to/Medium), HUNT (ProductHunt), SCAN (find communities), BLITZ (all at once). Use via /anima-marketing [MODE] [project-url-or-path]. Supports browser automation for X.com, Reddit, HN via native Computer Use (embedded) and MCP browser tools as fallback. GitHub submissions via gh CLI.
ANIMA company strategy war room. Use when discussing AIFLOW LABS goals, ANIMA roadmap, competitive positioning, fundraising, Shenzhen trip planning, module prioritization, hardware integration strategy, or any company-level decision making. Loads full brainstorming vault + ANIMA ecosystem memory + competitive intelligence. Modes: BRAINSTORM (open discussion), STATUS (where are we), PLAN (sprint planning), COMPETE (competitive analysis), FUND (fundraising prep), DECIDE (make + record decisions).
ANIMA Supervisor Agent — The safety-first, quality-enforcing agent that oversees all ANIMA module development, pipeline compilation, and deployment. Acts as the SDK's brain during development. Modes: BUILD (module development), VALIDATE (pipeline checking), SAFETY (supervisor design), INTEGRATE (hardware integration), DATA (flywheel pipeline). Use when building ANIMA modules, designing safety systems, validating pipelines, or working on the Intelligence Compiler SDK.
ANIMA Strategy Validator — Goes online to find pitfalls, validate claims, and stress-test the ANIMA strategy against what competitors are ACTUALLY building. Searches for real announcements, GitHub repos, funding news, product launches, and technical claims that could invalidate or strengthen our positioning. Modes: VALIDATE (check our claims), HUNT (find new competitors), DESTROY (build destruction playbook for specific competitor), MONITOR (track changes).
基于 SOC 职业分类
| name | autoresearch-compare |
| description | Compare autoresearch results across runs, branches, or machines. Reads results.tsv from multiple branches and generates comparison tables. Use when the user asks "compare runs", "how did machine X vs Y", or "cross-run analysis". |
Compare experiment results across different autoresearch runs.
git branch --list "autoresearch/*"results.tsv and final train.py config| Run | Machine | Experiments | Best val_bpb | Improvement | Key Changes |
|-----|---------|-------------|--------------|-------------|-------------|
| ... | ... | ... | ... | ... | ... |