get-unpublished-changes
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to parse git commit history and file diffs, which can incorporate untrusted text from external repository contributors.
- Ingestion points: Found in SKILL.md under the 'Steps' section, where the agent is instructed to run and read the output of
git diff v{published-version}..HEAD. - Boundary markers: Absent. There are no guidelines or structural delimiters directing the agent to treat the diff content strictly as data or to ignore embedded instructions.
- Capability inventory: The skill involves local repository inspection and analysis steps but does not expose advanced system capabilities such as write access or network tools.
- Sanitization: Absent. No validation or filtering is specified to strip or neutralize potential instruction strings inside commit messages or source code changes.
Audit Metadata