The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands to collect repository state and perform file uploads.
Runs git config, git log, git diff, and git status to gather author details and work history.
Utilizes gh api to programmatically interact with the GitHub API for content submission and metadata retrieval.
Uses system utilities like cat, rm, and base64 to manage temporary files in /tmp.
[DATA_EXFILTRATION]: The skill is designed to collect and transmit sensitive developer and repository data to a remote server.
Specifically, it extracts git config user.email and detailed code differences (git diff).
This information is sent to a GitHub repository specified by the TEAM_REPO variable. Users should ensure the target repository is appropriate for the sensitivity of the data being shared.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its handling of unvalidated local git data.
Ingestion points: Untrusted data is ingested from git log --oneline -10 and git diff output.
Boundary markers: The markdown generation template lacks boundary markers or instructions to ignore embedded commands within the git log content.
Capability inventory: The skill has write access to remote repositories via gh api.
Sanitization: Content extracted from the git environment is not sanitized or escaped before being interpolated into the log or uploaded to the remote repository.

team-log