The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell scripts to traverse the ~/Dev directory and runs git and gh commands to extract repository URLs and pull request metadata.
[DATA_EXFILTRATION]: Information regarding pull requests (titles, authors, and branches) is sent to an external Telegram relay service via the a2a_send tool. This is the documented purpose of the skill for the user 'Espen'.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from GitHub pull requests, which could contain malicious instructions.
Ingestion points: Data is ingested via the gh pr list command in SKILL.md.
Boundary markers: None; PR titles and branch names are interpolated directly into the status report and alert message.
Capability inventory: The skill has shell execution capabilities and access to the a2a_send tool for external communication.
Sanitization: While jq is used to parse the structured JSON from GitHub, there is no content sanitization or filtering applied to the resulting strings before they are processed by the agent.

pr-monitor