The Agent Skills Directory

[SAFE] (SAFE): Analysis of the provided Python templates and markdown documentation reveals no evidence of malicious intent or vulnerabilities.
[DATA_EXFILTRATION] (LOW): The skill utilizes a placeholder variable DATABASE_URL for Postgres connections. There are no hardcoded credentials, API keys, or unauthorized file system access patterns.
[INDIRECT_PROMPT_INJECTION] (LOW): While the skill assists in building agents that process user input, the templates themselves do not provide an unusual or unsafe attack surface beyond the inherent nature of the LangGraph framework.
[REMOTE_CODE_EXECUTION] (SAFE): No remote script downloads, dynamic code execution (eval/exec), or suspicious package installations were found.

checkpoint-patterns