omh-gateway-intent-card
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external communications, creating an ingestion surface for potentially untrusted data. 1. Ingestion points: User requests and session data from Discord, Slack, and Telegram messaging gateways. 2. Boundary markers: The skill instructions mandate a prepared-vs-observed boundary to ensure the agent distinguishes between its own guidance and verified platform evidence. 3. Capability inventory: The skill utilizes shell command execution via the omh CLI tool for status recording. 4. Sanitization: No explicit character-level sanitization is defined, relying instead on logical separation of intent and evidence.
- [COMMAND_EXECUTION]: The skill employs a platform-specific command-line interface (omh) to record runtime execution states and orchestration events.
Audit Metadata