The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions and the helper script (scripts/gh-env.sh) access standard sensitive file paths such as ~/.git-credentials and ~/.hermes/.env to retrieve GitHub authentication tokens. This behavior is necessary for the skill's stated purpose of managing and detecting GitHub authentication and does not involve unauthorized transmission of data.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection as it processes untrusted data from the GitHub API and local git remote configurations.\n
Ingestion points: Git remote URL detection (git remote get-url) and GitHub API responses (api.github.com/user) in scripts/gh-env.sh.\n
Boundary markers: None identified in the helper script.\n
Capability inventory: The script performs network operations using curl and executes shell commands including git and gh.\n
Sanitization: The script uses python3 to safely parse JSON data from the API, which mitigates simple injection attempts in the data stream.\n- [COMMAND_EXECUTION]: The skill provides numerous shell commands for the user or agent to configure git and the gh CLI, which is typical for an authentication setup utility.

github-auth