release-management
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill utilizes dynamic context injection (using the
!commandsyntax) inscripts/create-release.mdto automatically populate the agent's context with repository metadata, such as the current version, branch name, and recent commit history. These commands are limited to read-onlygitoperations and do not include user-supplied arguments, which prevents command injection at the loading phase.\n- [INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it reads git commit messages and PR bodies into the prompt context to generate release notes. While this is a standard and necessary part of release management, malicious instructions could theoretically be embedded in commit messages by an attacker with commit access to the repository.\n - Ingestion points: The
!git logcommand inscripts/create-release.mdfetches commit history.\n - Boundary markers: Commit messages are presented as a markdown list without specific delimiters or instructions to ignore embedded commands.\n
- Capability inventory: The skill has access to
Bash,Write, andEdittools, allowing for file system and repository modifications.\n - Sanitization: There is no sanitization of the commit message content before it is processed by the agent.
Audit Metadata