The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions to deliberately bypass local environment security controls. Specifically, it directs the agent to create a marker file (.commit-skill-active) within a temporary directory to suppress the execution of the git-commit-block hook. This represents a policy override pattern designed to circumvent established workflow constraints.
[COMMAND_EXECUTION]: The skill utilizes shell commands to interact with the repository's filesystem and git CLI. It performs directory creation (mkdir -p), file creation (touch), and deletion (rm -f) based on the output of command substitution ($(git rev-parse --show-toplevel)). While functional, these operations are used to implement the security hook bypass.
[INDIRECT_PROMPT_INJECTION]: The skill demonstrates an attack surface for indirect prompt injection (Category 8):
Ingestion points: The agent ingests untrusted data from the local repository using git diff --cached and git diff during the 'Analyze Changes' phase.
Boundary markers: There are no boundary markers or instructions to treat the ingested diff content as data rather than instructions.
Capability inventory: The skill possesses the capability to modify the repository state via git add and git commit commands.
Sanitization: No sanitization or escaping is performed on the content extracted from file diffs before it is interpolated into the prompt for commit message generation.

commit