The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its core workflow of analyzing external, untrusted data.
Ingestion points: The skill reads data from the local repository using git diff, git log, and the Read tool to examine file contents.
Boundary markers: No explicit boundary markers or delimiters are defined to separate the skill's instructions from the content of the files being analyzed.
Capability inventory: The skill possesses the capability to suggest executable shell commands and code snippets (e.g., node -e, npx jest) based on the analyzed content. While the tools listed in allowed-tools are restricted, a user might execute suggested malicious commands manually.
Sanitization: No sanitization or validation logic is present to filter out potential instructional overrides embedded in code comments or string literals within the repository.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to interact with the local environment. Although it is restricted to specific git subcommands (diff, log, status, show), these commands provide access to potentially sensitive repository metadata and history.

explore-test