skills/modem-dev/hunk/hunk-release/Gen Agent Trust Hub

hunk-release

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs extensive shell automation to manage the release lifecycle. This includes using git for tagging and branching, bun for building and benchmarking, npm for package verification, and the GitHub CLI (gh) for managing releases and monitoring CI workflows.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with well-known developer services, including npmjs.org for package distribution and GitHub for repository management and binary attestation. These operations are essential for the skill's primary purpose of publishing software releases.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an indirect injection surface because it reads and processes data from external files and tool outputs to generate documentation.
  • Ingestion points: The workflow reads content from .changeset markdown files, website/releases/notes.json, and the output of commands like npm view and gh run to build release notes and metadata.
  • Boundary markers: The skill instructions emphasize safety by requiring the agent to verify generated diffs and obtain explicit user confirmation before executing critical actions like pushing tags or publishing packages.
  • Capability inventory: The execution environment allows for broad file system access, network communication via standard developer tools, and arbitrary command execution through bun run scripts.
  • Sanitization: The instructions do not specify automated sanitization for the content of changesets or release note inputs, relying instead on manual review of the resulting diffs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 03:59 AM
Security Audit — agent-trust-hub — hunk-release