tinymist-release

Pass

Audited by Gen Agent Trust Hub on Apr 4, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local repository scripts, specifically scripts/release-preflight.mjs and scripts/link-docs.mjs, to gather release metadata and perform local file updates.
  • [COMMAND_EXECUTION]: The workflow incorporates automated execution of commands generated by the project's preflight script. These are restricted to local preparation tasks like patching manifests, updating changelogs, and staging git commits.
  • [SAFE]: A critical security boundary is enforced where the agent is strictly prohibited from executing commands with external side effects—including yarn release, cargo publish, and various gh (GitHub CLI) operations—without explicit, immediate maintainer approval.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes instructions and data from local repository files and script outputs.
  • Ingestion points: The JSON output of scripts/release-preflight.mjs and the contents of documentation files like docs/tinymist/release-instruction.typ.
  • Boundary markers: Present. The skill defines a clear 'External actions' phase that serves as a hard checkpoint for human verification.
  • Capability inventory: Includes shell command execution via Node.js, Git, Yarn, Cargo, and the GitHub CLI.
  • Sanitization: The skill relies on manual maintainer review of proposed commands rather than automated input sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 4, 2026, 08:33 AM
Security Audit — agent-trust-hub — tinymist-release