omh-sales-pipeline-review
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external CRM exports and pipeline snapshots, as documented in SKILL.md and references/procedure.md. This ingestion of untrusted data constitutes an attack surface for indirect prompt injection.
- Ingestion points: CRM exports and pipeline snapshots are provided by the user via JSON or manual input as defined in the
preparecommand and clarification questions. - Boundary markers: The workflow implements a strict
sales_pipeline_validate_scopestep that requires aHOLDdisposition if data quality, freshness, or definitions are not explicitly confirmed. - Capability inventory: The skill is restricted to analytical outputs and brief preparation using the
omhruntime tool; it explicitly lacks capabilities for data storage, CRM mutation, network communication, or outreach. - Sanitization: The procedure enforces strict schema adherence (e.g.,
sales_pipeline_scope/v1) and requires the agent to cite record references for all findings, preventing the interpolation of unverified data into executive summaries.
Audit Metadata