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

release

Pass

Audited by Gen Agent Trust Hub on Feb 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local commands and scripts to perform release tasks. Specifically, it calls a project-specific script ./scripts/bump-version.sh to update project metadata. It also utilizes git for branching, committing, tagging, and pushing, and the GitHub CLI (gh) to view pull requests/issues, create/merge PRs, and monitor CI runs.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external GitHub resources to build changelogs.
  • Ingestion points: The skill fetches author information and PR/issue data using gh pr view and gh issue view commands.
  • Boundary markers: No specific delimiters or instructions are provided to the agent to isolate or ignore potentially malicious content within the fetched data.
  • Capability inventory: The agent has the authority to perform high-impact actions, such as committing code, pushing tags, and merging pull requests, based on its interpretation of the processed data.
  • Sanitization: There is no evidence of sanitization or validation of the data retrieved from GitHub before it is incorporated into generated markdown files or PR descriptions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 26, 2026, 08:29 AM