caveman-commit
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it ingests untrusted data (git diffs) and uses it to perform persistent file-system modifications.
- Ingestion points: The skill implicitly reads git diffs or staged changes to generate commit messages and populate the "Output Report" section (SKILL.md).
- Boundary markers: Absent. The instructions do not specify any delimiters or warnings to ignore instructions that might be embedded within the source code or diffs being processed.
- Capability inventory: The skill performs multiple file-write and file-append operations. Specifically, it creates new files in
llmwiki/wiki/draft/cave/and appends content tollmwiki/wiki/index.mdandllmwiki/wiki/log.md(SKILL.md). - Sanitization: Absent. Data derived from the diffs (such as the
<ten>summary or the "What" section of the report) is not sanitized or escaped before being written to the filesystem.
Audit Metadata