The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill is designed to ingest and process untrusted log data which may contain malicious instructions meant to influence the agent's next actions.
Ingestion points: The scripts/parse-logs.py script reads external log files provided as command-line arguments.
Boundary markers: The postmortem template uses markdown code blocks (triple backticks) to separate log data from the report structure, which serves as a weak boundary.
Capability inventory: The Python script has local file-read capabilities via pathlib.Path.read_text().
Sanitization: No sanitization or escaping is performed on the log content before it is extracted into the JSON output or interpolated into the template.

debugging