skill-state-manager（技能状态管理器）

🛡️ Skill State Manager (SSM)

Say goodbye to manual config editing and endless repetitive token pasting!

The Skill State Manager (SSM) is a revolutionary "Meta Skill" designed for modern AI assistants (like Claude Code, Codex). It acts as your invisible, native auth butler. Whenever a third-party tool (like Notion, Jira, or a custom API) needs a secret or API key, SSM automatically steps in.

No more hunting down hidden .env files. No more hardcoded secrets. Just a seamless, native conversational experience—if a tool needs an auth token, your AI will simply ask you for it *once*, save it securely, and then immediately resume your workflow.

🌟 Why You'll Love It

• 🗣️ Native Conversational Flow: You never have to leave the chat. If a key is missing, the AI pauses the tool, politely asks you for the secret in the chat, and saves it.
• 🚫 Zero Config Setup: Stop writing boilerplate auth logic for every new tool you create. Let SSM handle the prompting and persistence.
• 🤝 Platform Aware: It automatically adapts to your CLI environment (Claude Code, Codex), storing states cleanly in ~/.ai-skills-state/.

🤖 Instructions for the AI Assistant

As an AI assistant, you are equipped with the "Skill State Manager" (SSM) rules to handle authentication and state persistence for custom third-party tools/skills.

WHEN TO TRIGGER SSM:

• DO NOT trigger for built-in tools (e.g., shell, file read/write, search) or public tools without authentication.
• TRIGGER SSM when you are about to use a custom tool that interacts with third-party authenticated services (e.g., Jira, Notion, Custom APIs) OR when a tool explicitly requires secrets.
• TRIGGER SSM if a tool execution fails with a 401/403 authentication error.

SSM WORKFLOW:

When triggered, you MUST follow these exact steps sequentially:

1. Detect Platform: Determine which AI platform you are currently running in. Your behavior must adapt based on this detection:

- Claude Code: Uses a global storage approach (~/.claude/skills-state/). - Codex / Other CLI: Uses a generic global storage approach (~/.ai-skills-state/).

1. Check Storage: Look for the credentials in the designated local directory based on the detected platform:

- If Claude Code: check ~/.claude/skills-state/<tool-name>.json - If Codex / Other: check ~/.ai-skills-state/<tool-name>.json

1. Prompt User (If Missing):

- If the required credentials (e.g., API keys, tokens) are not found in the storage, STOP the tool execution. - Ask the user politely for the required information via a chat message. - *Example*: "The current operation requires the Notion API, but your API key was not found. Could you please provide it? I will save it locally in ~/.claude/skills-state/notion-skill.json (or the appropriate path) for future use."

1. Save State:

- Once the user provides the information in the chat, use the Write tool to save it into the JSON file resolved in Step 1. - For Claude Code, write to ~/.claude/skills-state/<tool-name>.json. - For Codex / Other CLI, write to ~/.ai-skills-state/<tool-name>.json. - Ensure the content is stored as valid JSON. - Proceed with the actual tool execution using the newly acquired credentials.

🛠️ Usage Example

User: "Can you query my Jira board for high-priority bugs?" AI (Internal Thought): *I need to use the Jira tool. Let me check ~/.ai-skills-state/jira-tool.json.* AI (Internal Thought): *File not found. Triggering SSM.* AI: "To query your Jira board, I need your Jira API token and URL. Could you please provide them here? I'll save them locally in ~/.ai-skills-state/jira-tool.json so you won't have to enter them again." User: "Sure, my token is xyz123 and url is https://myorg.atlassian.net." AI (Internal Thought): *Writes data to ~/.ai-skills-state/jira-tool.json using the Write tool.* AI: "Thanks! I've saved your credentials. Now, let's look at those high-priority bugs..." *(Proceeds to execute Jira query)*