The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run within collect_feedback_diagnostics.py to execute the moviepilot doctor command. This is used for gathering system health metadata to include in the bug report. The execution uses a fixed list of arguments and does not involve a shell, mitigating command injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from log files and user error reports. It implements defensive measures by explicitly instructing the agent to ignore any embedded instructions in logs or pasted text. It also uses a comprehensive redact_logs function that uses regular expressions to scrub sensitive information (tokens, cookies, IPs, passwords) from the ingested data before it is processed or presented to the user.
[DATA_EXPOSURE]: The skill accesses local application logs and configuration files. However, this is gated by explicit user intent and includes a mandatory preview step where the user must review the redacted logs and confirm the target repository before any data is exfiltrated to GitHub.
[EXTERNAL_DOWNLOADS]: No external code or packages are downloaded at runtime; the skill relies on the existing environment and pre-installed dependencies of the MoviePilot application.

feedback-issue