The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted content from the repository (e.g., code diffs, markdown files) which could contain hidden instructions for the agent.
Ingestion points: The agent ingests git diff --staged and repository files like .gitignore and README.md as described in SKILL.md.
Boundary markers: The instructions lack explicit boundary markers or directions to ignore instructions found within the audited data.
Capability inventory: The skill can modify repository files (.gitignore, README.md, GRAPH_REPORT.md) and execute external commands (git, Graphify).
Sanitization: There is no evidence of sanitization for the external data processed by the agent.- [COMMAND_EXECUTION]: The skill relies on executing system commands and external tools to perform its auditing tasks.
Evidence: SKILL.md instructs the agent to run git commands and the Graphify utility for cycle detection and path verification.

devops-git-guard