The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands using git and the GitHub CLI (gh) to manage project history and issue tracking. This includes actions such as git log, git clone, git push, and gh issue create, which are standard operations for a development-focused agent.
[DATA_EXFILTRATION]: The skill involves network communication by cloning the repository from the origin and pushing updates back to it. These network operations are directed at the established project repository and are used for synchronizing documentation, not for unauthorized data removal.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it reads and processes data from external sources such as GitHub issue titles and states. 1. Ingestion points: External data is ingested via gh issue view, gh issue list, and by reading the contents of MEMORY.md from the remote repository. 2. Boundary markers: There are no explicit markers or instructions provided to the agent to treat this ingested data as potentially untrusted. 3. Capability inventory: The skill can modify the remote repository via git push and create new issues via gh issue create. 4. Sanitization: No sanitization or filtering logic is present to clean the ingested data before it is processed by the agent.

reflect