The Agent Skills Directory

[SAFE]: The skill is designed for text transformation and local logging. Its file-writing capabilities are restricted to the llmwiki/ directory, which is a standard pattern for documentation and logging.
[COMMAND_EXECUTION]: The skill documentation explicitly instructs the agent not to run commands like git commit or stage files, limiting its activity to generating text output.
[PROMPT_INJECTION]: The skill ingests untrusted data from git diffs to generate messages and reports, creating an indirect prompt injection surface. However, the impact is low as the only capability is writing text to a specific local path.
Ingestion points: Git diffs and project metadata used to generate commit messages and wiki logs.
Boundary markers: Absent; the skill does not explicitly tell the agent to ignore instructions embedded in the diffs.
Capability inventory: File-write operations specifically targeting the llmwiki/ folder.
Sanitization: Absent; the skill does not specify validation for the content of the diffs or the generated summaries.

caveman-commit