gomtm-agent-workflow
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process data from external, untrusted sources which creates a potential surface for indirect prompt injection.
- Ingestion points: Data enters the agent context via GitHub Issues, Telegram messages, and GitHub comments as described in the 'Source of Truth Split' and 'When to Use' sections of
SKILL.md. - Boundary markers: The skill implements 'Approval Gate Rules' in
SKILL.mdand specifies structured response templates inreferences/github-issue-kanban-approval.mdto help distinguish commands from general text. - Capability inventory: The workflow utilizes the
hermesCLI tool for Kanban management (hermes kanban create/boards) and interacts with the local file system for workspace management. - Sanitization: The skill includes a manual verification step in
references/github-issue-kanban-approval.md('确认评论来源可信' - verify comment source is trusted) and requires explicit approval keywords ('批准执行') before proceeding with tasks.
Audit Metadata