The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs a shell command to be executed in WSL by interpolating user-provided $ARGUMENTS into the {flags} placeholder. If the agent does not strictly validate the input against the provided scope table, an attacker could supply arguments containing shell metacharacters to execute arbitrary commands within the WSL environment.
[INDIRECT_PROMPT_INJECTION]: The skill implements an automated 'Self-Healing Loop' that parses results from the external CodeRabbit CLI and attempts to automatically modify the codebase based on detected 'CRITICAL' findings.
Ingestion points: CodeRabbit CLI findings parsed in Step 4 of SKILL.md.
Boundary markers: Absent; the instructions do not include delimiters or warnings to ignore potentially malicious instructions embedded in review findings.
Capability inventory: File-write access (for 'auto-fix' iterations) and shell execution (to run and re-run the CodeRabbit tool) as described in Steps 2 and 5.
Sanitization: Absent; the skill lacks validation or filtering of the content returned by the external tool before using it to drive code changes.
[EXTERNAL_DOWNLOADS]: The error handling section suggests installing the coderabbit-cli package from the official Python package registry (PyPI) if the command is not found in the WSL environment.

coderabbit-review