skills/chann/skills/build-reinstall/Gen Agent Trust Hub

build-reinstall

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements strong security boundaries by requiring all build and reinstall commands to be discovered from local project evidence (e.g., .build-reinstall.yaml, package.json, Makefile) rather than using guessed or hardcoded defaults.
  • [SAFE]: It explicitly forbids dangerous operations such as sudo, privilege escalation, network exfiltration, or modifications outside the specific project root and resolved install targets.
  • [SAFE]: The workflow prevents automatic execution by requiring an explicit user request, and it mandate a 'preflight' display of the plan (commands, paths, outputs) before any mutations occur.
  • [SAFE]: Integrity verification is enforced by comparing SHA-256 digests of built versus installed artifacts, and failure at any step (build, resolve, or install) immediately halts the process to protect the system state.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 02:04 PM
Security Audit — agent-trust-hub — build-reinstall