04-multi-platform-publish
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a mandatory 'Dry Run and Explicit Authorization' step, ensuring no automated publishing occurs without a human reviewing the complete plan first.\n- [SAFE]: An idempotency check is performed using a local ledger to prevent redundant or unauthorized republishing of assets that have already been distributed.\n- [SAFE]: Capability resolution includes specific checks for integration gaps in third-party providers (e.g., Composio LinkedIn/TikTok limitations), ensuring the agent does not attempt unsupported actions.\n- [SAFE]: Disclosure and jurisdictional policy gates are consistently applied by calling a prerequisite guardrails skill, preventing accidental non-compliance with platform rules.\n- [SAFE]: Supporting scripts (eval/contract_check.py and eval/run-eval.sh) are limited to local validation and testing tasks, containing no malicious code or external network dependencies.
Audit Metadata