The Agent Skills Directory

[CREDENTIALS_UNSAFE] (MEDIUM): The script scripts/extract.sh searches the user's home directory (~/.mcp-auth/) for credential files (*_tokens.json) to extract authentication tokens. While necessary for the skill's intended MCP integration, this involves broad access to sensitive credential storage.\n- [EXTERNAL_DOWNLOADS] (MEDIUM): The script uses npx -y mcp-remote to initiate an OAuth flow. This command downloads and executes the mcp-remote package from the npm registry at runtime without a pinned version, creating a potential supply chain risk.\n- [PROMPT_INJECTION] (LOW): This skill is vulnerable to indirect prompt injection as it ingests untrusted content from external URLs. \n
Ingestion points: External URL content is fetched via the Tavily API in scripts/extract.sh.\n
Boundary markers: Absent; the extracted content is returned without delimiters.\n
Capability inventory: Execution of shell commands (curl, jq, npx) and background processes.\n
Sanitization: None; the raw content from the API is passed back to the agent context.\n- [COMMAND_EXECUTION] (LOW): The script executes shell commands to perform API requests and process JSON data. It interpolates user-provided URL strings into its logic through jq and curl.

extract