omh-deep-interview
Pass
Audited by Gen Agent Trust Hub on May 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of ingesting and synthesizing external data into project specifications.
- Ingestion points: The agent ingests untrusted data from user conversation responses in Phase 2 and confirmed research reports located at
.omh/research/*-report.mdin Phase 0 (SKILL.md). - Boundary markers: The instructions do not define explicit boundary markers or delimiters to isolate user-provided content from the agent's logic, nor do they instruct the agent to ignore instructions embedded within the research reports.
- Capability inventory: The skill uses the
omh_statetool for reading and writing state, performs file enumeration via the terminal, and writes generated specifications to the local file system (SKILL.md, Phase 0, 3, and 5). - Sanitization: There is no evidence of sanitization, escaping, or validation of the ingested external content before it is interpolated into the final specification template.
Audit Metadata