The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that override the agent's default tool selection behavior by directing it to use Nango for all external API calls, even when other tools or MCPs might be available.
Evidence: "TRIGGER on any request to read or modify data in an external product, even when no matching MCP tool is loaded" and "Do not use MCPs, provider CLIs, or ad-hoc direct API calls outside Nango."
[DATA_EXFILTRATION]: The skill is designed to route sensitive data from various third-party services (e.g., Google Calendar, Gmail, Slack, Notion) through Nango's external infrastructure.
Evidence: The skill instructs the agent to use https://api.nango.dev (or a configured NANGO_SERVER_URL) for all reading and modifying of data in external products.
[CREDENTIALS_UNSAFE]: The skill requires a secret key and provides instructions for users to store it in environment variables or configuration files.
Evidence: The instructions guide the user to set NANGO_TOOLBOX_SECRET_KEY in a local .env file or shell profile.
[PERSISTENCE]: The skill attempts to maintain access to its required credentials across sessions by modifying the user's environment configuration.
Evidence: The instructions suggest setting the secret key in the user's shell profile and offer to add it automatically.
[EXTERNAL_DOWNLOADS]: The skill suggests installing additional tools from a remote repository.
Evidence: It references installing the building-nango-functions-remotely skill from https://github.com/NangoHQ/skills, which is the vendor's official repository.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external APIs which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Data is ingested from third-party APIs such as Gmail, Slack, Notion, and HubSpot via Nango Proxy or Actions.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are provided for the data returned from these external services.
Capability inventory: The skill has the capability to perform network operations (GET, POST, PATCH) to read and modify data on external platforms.
Sanitization: There are no explicit instructions for sanitizing or validating the content retrieved from external APIs before processing.

nango-toolbox