objectstack-automation
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of automation that processes data from potentially untrusted external sources, creating a surface for indirect instructions to influence agent behavior.
- Ingestion points: Inbound HTTP webhooks (api flows) and record change events (record_change flows) defined in SKILL.md.
- Boundary markers: Data interpolation uses CEL envelopes and template holes for delimited evaluation.
- Capability inventory: Includes HTTP requests, database CRUD operations, messaging, and custom script execution as documented in SKILL.md.
- Sanitization: The framework relies on os validate for compile-time schema and expression verification.
- [PRIVILEGE_ESCALATION]: The skill documents mechanisms for elevating execution privileges to perform administrative tasks or bypass standard record restrictions.
- Evidence: SKILL.md provides instructions on using runAs: 'system' for writing read-only fields and warns about the risks of unscoped multi-tenant data access when performing system-level sweeps.
Audit Metadata