The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Git commands to perform branch synchronization tasks such as fetching, merging, rebasing, and stashing. It implements safety protocols including the use of --force-with-lease to prevent overwriting remote work and strictly forbids force-pushing to the main or master branches.
[DATA_EXFILTRATION]: Analyzes local repository metadata, including branch history, status, and remote configurations. These operations are standard for Git-based workflows and do not involve transmitting sensitive data to external or untrusted destinations.
[PROMPT_INJECTION]: Includes 'Anti-Hallucination Guidelines' and specific workflow phases that require the agent to verify the actual state of the repository before suggesting or executing actions, which serves to reinforce operational safety.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from branch names, commit messages, and PR metadata. While this presents a potential surface for indirect injection, the risk is mitigated by instructions that mandate manual conflict resolution and user confirmation for high-risk operations like force-pushing.

git-sync