菜单
Store user preferences, learnings from tasks, and procedure outcomes. Use to remember what works and recall context before new tasks. (user)
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
Disciplined development workflow for AI coding agents working in the Atulya monorepo — covers context gathering, branch hygiene, edit/lint/test cycle, API surface changes that require client regen, parallel vs sequential tool use, and the report-back template. Use at the start of any non-trivial Atulya coding task, or when the user asks how to "implement X", "port Y", "change the API", or "ship a feature" in this repo.
Run the Atulya repo through end-to-end system testing — toolchain check, lint, Python pytest, Rust cargo check, TypeScript build, control-plane typecheck, OpenAPI + client regen as a contract test, cross-client field smoke test, and legacy-leak audit. Use when validating a change before commit/PR, after pulling main, before a release, or when the user asks to "run all tests", "verify everything", "system sanity check", or "end-to-end test".
Design a meaningful git commit for the Atulya repo end-to-end — what commands to run to inspect the change, how to decide if it's one commit or several, how to write a message in this repo's `[type] ~ subject` convention with a structured body grouped by surface, and the exact `git add` / heredoc `git commit` invocations to use. Use whenever the user says "commit this", "git commit", "draft a commit message", "what should the commit be", or you've just finished a feature/fix and want to propose a commit before they ask.
Complete Atulya documentation for AI agents. Use this to learn about Atulya architecture, APIs, configuration, and best practices.
Atulya Cortex (TUI/WhatsApp), Hand tools, internet stack (SearXNG/Firecrawl), config, and what-if troubleshooting. Use for cortex wiring, web_search/web_extract, deliberation limits, or docker internet-search compose.
Review changed code against project standards. Checks for missing tests, dead code, type safety, lint issues, and coding conventions. Run after completing any implementation work.
| name | atulya-local |
| description | Store user preferences, learnings from tasks, and procedure outcomes. Use to remember what works and recall context before new tasks. (user) |
You have persistent memory via the atulya-embed CLI. Proactively store learnings and recall context to provide better assistance.
Before using memory commands, verify Atulya is configured:
uvx atulya-embed daemon status
If this fails or shows "not configured", run the interactive setup:
uvx atulya-embed configure
This will prompt for an LLM provider and API key. After setup, the commands below will work.
Use memory retain to store what you learn:
uvx atulya-embed memory retain default "User prefers TypeScript with strict mode"
uvx atulya-embed memory retain default "Running tests requires NODE_ENV=test" --context procedures
uvx atulya-embed memory retain default "Build failed when using Node 18, works with Node 20" --context learnings
Use memory recall BEFORE starting tasks to get relevant context:
uvx atulya-embed memory recall default "user preferences for this project"
uvx atulya-embed memory recall default "what issues have we encountered before"
Use memory reflect to synthesize context:
uvx atulya-embed memory reflect default "How should I approach this task based on past experience?"
Always store after you learn something valuable:
Always recall before: