The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bash script (scripts/sync.sh) to perform git operations. It also provides the agent with specific shell commands to manage git worktrees and resolve merge conflicts manually, including git merge, git worktree remove, and git branch -d. It further manages system state by removing the .git/index.lock file if it is determined to be stale.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from a remote git repository during synchronization.
Ingestion points: The git pull and git fetch commands within scripts/sync.sh fetch external content into the local environment.
Boundary markers: Absent. There are no instructions to the agent to treat the pulled content or resolved conflicts as potentially untrusted or to ignore embedded instructions.
Capability inventory: The agent has the capability to execute shell scripts (bash), modify the file system via git commands and rm, and perform network operations (git push).
Sanitization: Absent. The skill's conflict resolution logic instructs the agent to automatically merge content (e.g., 'accept both'), which could allow malicious instructions from the remote source to persist in the workspace.

sync-git