菜单
Fast token-efficient execution loop for active repo work. Useful for Project X / AG Trade style workflows where agents must stay scoped, safe, and brief.
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。
基于 SOC 职业分类
Compact Conventional Commit generator. Use for commit messages, changelog snippets, and staged-change summaries.
Safe compression for natural-language memory files like AGENTS.md, CLAUDE.md, notes, todos, and preferences. Creates .slate.md output by default.
Compact debugging workflow. Use for failed tests, broken builds, runtime errors, crashes, dependency issues, and unexpected behavior.
Compact patch summary. Use after Codex or Claude Code edits files and the user needs a fast handoff.
Sentence-level language minifier. Removes linguistic slack from agent output while preserving code, commands, errors, and safety context.
Terse PR/code review comments. Use for pull requests, diffs, file reviews, and review-comment drafting.
| name | slate-run |
| description | Fast token-efficient execution loop for active repo work. Useful for Project X / AG Trade style workflows where agents must stay scoped, safe, and brief. |
Fast execution loop for real projects.
Maximize useful progress per token during active repo work.
Use for Project X, AG Trade, repo maintenance, bugfix loops, dashboard edits, risk-control changes, and any task where the agent should avoid broad scanning or long narration.
Goal:
Files:
Action:
Check:
Result:
Next:
For trading, payments, auth, secrets, deploys, destructive DB ops, or real capital:
Goal: add rejected-trade journal tile.
Files: dashboard.py, btc5m_journal.py.
Action: added rejected_count + reason summary.
Check: `python cli.py verify` passed.
Result: dashboard shows fills + rejects.
Next: add reject-rate trend chart.