The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines several shell command templates for system diagnostics, including log inspection (tail, journalctl, rg), git history analysis (git log, git blame), and dependency checking (npm, pip, go). These commands are intended for manual or automated execution during an investigation and align with the skill's stated purpose.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection due to its core function of analyzing external data.
Ingestion points: Processes application logs (/var/log/app.log), system journal entries, and build output via references/techniques.md.
Boundary markers: No specific delimiters or safety instructions are defined to separate untrusted log content from the agent's instructions.
Capability inventory: Subprocess execution for file reading, searching, and version control operations.
Sanitization: No content filtering or sanitization is implemented for the ingested logs or error messages.
[DATA_EXPOSURE]: The skill suggests commands to inspect environment variables (printenv) and configuration files (cat config.yaml). While these files often contain sensitive information, their inspection is a routine part of the 5-Why investigation workflow, and the skill does not include any automated exfiltration mechanisms.

root-cause-investigator