skills/toy-crane/skills/self-ship/Gen Agent Trust Hub

self-ship

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon data from specification folders, which creates a surface for instructions embedded in those files to influence implementation and merge behavior.
  • Ingestion points: Reads from docs/specs/<slug>/, AGENTS.md, and CLAUDE.md to guide the implementation process.
  • Boundary markers: Explicitly instructs the agent to stop and ask the user for clarification if the approved meaning of a specification changes or if ambiguities arise.
  • Capability inventory: The skill can merge pull requests, perform git pushes with --force-with-lease, and execute local verification tools such as applications, browsers, or simulators.
  • Sanitization: No explicit technical sanitization is described; the skill relies on instructional constraints and human-in-the-loop confirmation for high-impact decisions.
  • [COMMAND_EXECUTION]: The skill requests authority to execute sensitive shell-level operations as part of its automated workflow.
  • Evidence: Granted authority to merge pull requests, force-push to specific work branches, and run verification environments like browsers or simulators.
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing dependencies from a remote source.
  • Evidence: Recommends using npx skills@latest add toy-crane/skills to install missing required components such as babysit-specs, implement, and merge. These resources are associated with the skill's author.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 07:14 AM