building-tools
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The documentation defines tools that ingest untrusted data from external sources via a configurable input schema. (SKILL.md, references/schemas/input.yml)\n
- Ingestion Points: External data enters the system through the
inputstage, defined by a JSON Schema.\n - Boundary Markers: The platform uses JSON Schema validation to enforce data structure; missing required properties or failed patterns result in 422 validation errors.\n
- Capability Inventory: Tools execute
pageProcessors(exports and imports) that can perform lookups or actions in external systems, potentially using data derived from the input.\n - Sanitization: Input data is validated against the root JSON Schema before processing.\n- [DYNAMIC_EXECUTION]: The skill describes the use of JavaScript hooks (
preMap,postMap,postResponseMap) and script-based transformations for custom logic. This enables the execution of programmable logic within the integration platform. (SKILL.md, references/common-schemas/transform.yml)
Audit Metadata