The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a sequence of shell commands to manage code changes, including Git operations such as status, diff, log, and commit, as well as project-specific validation using bun run check.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository environment to inform its actions and generate user-facing content, which presents a surface for potential instruction injection.\n
Ingestion points: Reads local configuration from package.json and version control output from git diff and git log.\n
Boundary markers: The instructions specify using HEREDOC syntax for the git commit command to separate the generated message from the shell command structure.\n
Capability inventory: The agent can execute arbitrary local scripts defined in package.json and perform file staging and commit operations.\n
Sanitization: No explicit sanitization or filtering of the ingested diff content is specified before it is processed by the model to generate outcomes.

finish-work