gtm-icp

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill implements a robust 'Link safety' protocol in references/contract.md, instructing the agent to treat URLs containing credentials, tokens, or session identifiers as unsafe and prohibiting their use or persistence.- [COMMAND_EXECUTION]: Persistence is handled through standard Git operations (add, commit, push, pull). These commands are strictly limited to the icps/ directory and are governed by a mandatory user acceptance loop that previews every file change before execution.- [SAFE]: File system access is narrowly scoped to specific organization nodes and their corresponding icps/ subdirectories. The instructions in references/contract.md explicitly forbid scanning .git contents or using broad recursive reads, which protects against unintended data exposure.- [SAFE]: The skill includes clear boundaries for workspace resolution, using a specific discovery path (~/.gtm/) and validation criteria (presence of ORG.md) to ensure the agent only interacts with intended project repositories.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 06:29 PM
Security Audit — agent-trust-hub — gtm-icp