The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages and processes sensitive Feishu API credentials, including FEISHU_APP_ID, FEISHU_APP_SECRET, and User Access Tokens. These are stored and retrieved from local configuration files such as .claude/feishu-config.env and feishu-token.json, granting the skill high-privilege access to the user's Feishu workspace.
[COMMAND_EXECUTION]: The orchestrator component (orchestrator.py) utilizes the subprocess module to programmatically execute multiple Python scripts within the skill's directory structure to manage the workflow.
[COMMAND_EXECUTION]: The verification sub-skill (doc_verifier.py) employs the Playwright library to launch and control a headless Chromium browser instance. This allows for automated navigation and interaction with Feishu URLs to verify document creation.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface when processing external Markdown data, characterized by the following: 1. Ingestion points: The md_parser.py script reads local Markdown files. 2. Boundary markers: There are no explicit markers or instructions provided to the agent to distinguish between its own logic and instructions embedded within the processed Markdown. 3. Capability inventory: The skill has the capability to perform network requests (requests) and execute system processes (subprocess). 4. Sanitization: The skill performs defensive cleaning of zero-width characters and specific Markdown markers in md_parser.py, which mitigates some obfuscation risks but does not filter for malicious instructional content.

feishu-doc-orchestrator