The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes pnpm run dev to initiate code reviews based on Git diffs. This involves running project-specific scripts to compare branches, commits, or uncommitted changes.
[PROMPT_INJECTION]: The skill processes untrusted data which creates a surface for indirect prompt injection. 1. Ingestion points: The skill reads review comments from command output (stdout) and contents from Git diffs. 2. Boundary markers: No specific delimiters or boundary markers are defined for the processed data. 3. Capability inventory: The agent can execute local commands via pnpm. 4. Sanitization: The skill provides specific instructions to the agent to avoid copying secrets, tokens, or credentials from the code diffs into the review comment system.

difit-dev