The Agent Skills Directory

[SAFE]: The skill relies on local file operations within the _project_specs/ directory and standard version control commands (git pull, git push, git commit). These are routine development activities.
[INDIRECT_PROMPT_INJECTION]: The skill defines a pattern where the agent reads and interprets content from repository files that could be modified by other users or processes. This creates a potential surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The agent is instructed to read _project_specs/team/state.md, _project_specs/team/contributors.md, and other files in the _project_specs/ directory.
Boundary markers: Absent. The agent reads the raw Markdown content.
Capability inventory: The skill utilizes cat, grep, and git for state management and synchronization.
Sanitization: Absent. The agent processes the file content as instructional data for reporting team status.
[COMMAND_EXECUTION]: Provides an example Git pre-push hook using git show and md5. This is a common practice for ensuring local state matches remote state and does not involve risky execution of untrusted code.

team-coordination