changelog
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local bash scripts (
draft.sh) and system commands (git,git-cliff) to extract commit history and determine version tags. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from git commit messages to generate changelog entries.
- Ingestion points: Commit messages retrieved via
git-cliffand version information frompackage.json(SKILL.md, draft.sh). - Boundary markers: The skill mandates a manual curation step (Step 5 in SKILL.md) where the agent presents the drafted entries to the user for confirmation or editing before writing any files.
- Capability inventory: The skill uses shell execution for data retrieval and the
Edittool for filesystem writes (SKILL.md). - Sanitization: Relies on the user-in-the-loop curation process to filter or rephrase potentially malicious or inaccurate content from commit messages.
Audit Metadata