release
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via the
.agents/release.override.skill.mdfile and git commit history. The override file is designed to take precedence over the skill's instructions, allowing for potential hijack of the release process. - Ingestion points:
.agents/release.override.skill.mdandgit logoutput. - Boundary markers: Absent; the skill explicitly gives precedence to the override file.
- Capability inventory: Full
Bashaccess, filesystemWrite/Edit, andgit pushoperations. - Sanitization: Absent; no validation is performed on the override file or commit messages.
- [COMMAND_EXECUTION]: The skill uses the
Bashtool to perform git operations using strings sourced from the repository (tags, branch names, file paths). This presents a risk of command injection if the repository contains maliciously crafted file names or git metadata designed to exploit shell syntax.
Audit Metadata