The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs the agent to verify environment setup by executing 'echo $LANGSMITH_API_KEY', which prints sensitive credentials directly to the terminal output where they may be captured in execution logs, shared history, or user interfaces.
[PERSISTENCE]: In the troubleshooting section, the skill directs the agent to append export commands for sensitive environment variables (LANGSMITH_API_KEY, LANGSMITH_PROJECT) to shell profile files such as '~~/.bashrc' or '~~/.zshrc' to ensure they persist across sessions, representing a significant persistence mechanism for sensitive data.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the 'langsmith-fetch' package from the official Python Package Index (PyPI). This utility is managed by a well-known service provider.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external execution traces that may contain untrusted content.
Ingestion points: Trace and thread data are retrieved via the 'langsmith-fetch traces' and 'langsmith-fetch threads' commands in SKILL.md.
Boundary markers: The instructions do not include specific delimiters or directives to the agent to ignore instructions embedded within the fetched trace data.
Capability inventory: The skill allows the agent to create directories ('mkdir'), filter data ('grep'), and write files to the local disk using shell redirection ('recent-traces.json').
Sanitization: The fetched trace data is not validated or sanitized before being written to files or analyzed by the agent.
[COMMAND_EXECUTION]: The skill relies on various shell commands including 'mkdir' for session folder management, 'grep' for error detection in JSON logs, 'pip' for dependency installation, and shell redirection for logging trace activity.

langsmith-fetch