fluid-release

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • Automated Command Execution: The skill uses shell commands for tasks like updating package versions and pushing git branches (e.g., npm pkg set and git push). These operations are necessary for managing the release process within the project repository.
  • Indirect Prompt Injection Surface: The agent's logic is guided by external inputs such as the release-schedule.md file and current repository state (git tags and pull request lists).
  • Ingestion points: Data enters the context from the release schedule file and the output of git and GitHub CLI commands.
  • Boundary markers: No explicit delimiters are defined for the ingested metadata.
  • Capability inventory: The skill can execute subprocesses, perform git pushes, and create pull requests or issues.
  • Sanitization: The skill does not explicitly mention sanitization of the retrieved metadata before using it to determine version numbers or branch names.
  • Autonomous Mode Capabilities: When operating in CI or autonomous mode, the skill performs actions like creating branches and pull requests without manual confirmation. This functionality is intended for use by authorized project maintainers.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 08:13 PM
Security Audit — agent-trust-hub — fluid-release