The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides various shell commands for setting up a local development environment, including generating encryption keys via openssl rand -base64 32, managing Docker containers with docker compose, and running microservices through npm scripts.
[DYNAMIC_EXECUTION]: A Python script is provided to handle log capture from the API server. This script uses the pty and os modules to fork a process and execute npx tsx within a pseudo-terminal to bypass output buffering, writing the results to /tmp/nango-srv.log.
[CREDENTIALS_UNSAFE]: The documentation contains hardcoded default credentials for the local PostgreSQL development environment (user: nango, pass: nango). While these are common for local dev setups, they are flagged as hardcoded credentials.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by instructing the agent to query the local database via docker exec and use the resulting output (e.g., user email or verification status) to determine subsequent browser-based actions.
Ingestion points: The output of the docker exec nango-db psql command in SKILL.md is ingested into the agent context.
Boundary markers: No specific delimiters or warnings are used to differentiate database output from instructional content.
Capability inventory: The agent can execute shell commands (npm, docker, python3) and perform browser navigation/interaction.
Sanitization: No sanitization or validation of the database output is performed before the agent processes it.

running-and-testing-locally