The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local bash script scripts/detect-forge-cli.sh to identify the repository hosting service. It also invokes system binaries such as git, gh (GitHub CLI), and glab (GitLab CLI) to perform branch pushes and PR/MR management tasks.
[DATA_EXFILTRATION]: The skill transmits repository content, including code diffs and commit messages, to external services (GitHub or GitLab). This is the intended purpose of the skill for creating pull or merge requests.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection, as it reads and summarizes untrusted data from the repository (e.g., commit messages and file diffs) to generate PR/MR descriptions.
Ingestion points: Repository metadata is retrieved via git log, git diff, and forge-specific view commands as defined in SKILL.md.
Boundary markers: The instructions lack explicit delimiters or warnings to ignore malicious instructions that might be embedded in commit history or diffs.
Capability inventory: The agent possesses capabilities to push code to remote repositories and modify review request metadata.
Sanitization: No validation or sanitization is performed on the repository content before it is processed by the agent to generate textual summaries.

pr