The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The documentation includes hardcoded default credentials for the local PostgreSQL database service (user: nango, pass: nango). While these are intended for a local development environment, hardcoded credentials pose a security risk if the environment is exposed.
[DYNAMIC_EXECUTION]: The skill provides a multi-line Python script executed via python3 -c to wrap the server process in a pseudo-TTY (pty.fork()) for log redirection to /tmp/nango-srv.log. This pattern involving runtime script generation and process management is a medium-risk execution vector.
[INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by instructing the agent to ingest and parse untrusted data from local application logs and database outputs to find sensitive signup verification tokens.
Ingestion points: The agent reads content from /tmp/nango-srv.log and the output of docker exec ... psql commands.
Boundary markers: None. The agent is instructed to search for patterns like signup/verification within the raw output.
Capability inventory: The skill utilizes shell execution, file system writes (to /tmp/nango-srv.log), database access, and external process management via Python.
Sanitization: No sanitization or validation of the ingested log data is performed before the agent processes it.

running-and-testing-locally