skills/manaflow-ai/cmux/cmux-release/Gen Agent Trust Hub

cmux-release

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local repository scripts ./scripts/bump-version.sh and ./scripts/release-pretag-guard.sh for version management and pre-tag validation. This is standard automation for release workflows and utilizes local project assets.
  • [PROMPT_INJECTION]: The skill processes untrusted data by reading git commit logs and CHANGELOG.md to update release notes, creating an indirect prompt injection surface. (1) Ingestion points: git commit history and CHANGELOG.md. (2) Boundary markers: Absent. (3) Capability inventory: Shell execution via git, gh, and local scripts. (4) Sanitization: No sanitization of ingested commit messages or changelog content is specified.
  • [SAFE]: References to the manaflow-ai/cmux GitHub repository and the list of required Apple signing secrets are legitimate vendor-specific resources and standard requirements for automated macOS software releases.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 09:59 PM
Security Audit — agent-trust-hub — cmux-release