The Agent Skills Directory

[COMMAND_EXECUTION]: The skill attempts to execute a script located at a hardcoded absolute filesystem path: /Users/phaedrus/Projects/prompts/skills/autoreview/scripts/autoreview. Relying on absolute paths in a specific user's home directory is an unsafe practice as the script's content is unverifiable and outside the skill's scope.
[REMOTE_CODE_EXECUTION]: The execution of the autoreview script represents an unverified code execution vulnerability since the script is not bundled with the skill and originates from an external local source.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted data from pull request titles, bodies, and issue comments, using them to construct search queries.
Ingestion points: Metadata and content from pull requests and issues fetched via gh commands in SKILL.md.
Boundary markers: Absent. The skill does not wrap external data in delimiters or provide instructions to disregard embedded commands.
Capability inventory: Shell execution capabilities via gh, git, and the absolute-pathed autoreview script.
Sanitization: Absent. Content from external sources is interpolated into search strings without sanitization or validation.

land-pr