building-flows
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill documents the use of custom JavaScript functions for data processing and routing within flows. This includes 'postResponseMap' hooks and script-based routers that execute logic at runtime based on script identifiers. Evidence is found in SKILL.md and the router.yml and page-processor.yml schemas.- [INDIRECT_PROMPT_INJECTION]: The skill defines a large surface for ingesting untrusted data from external sources such as webhooks, listeners, and remote file systems (FTP/S3). This creates a vulnerability where malicious instructions embedded in source data could influence agent behavior if the agent is used to process or debug live records.
- Ingestion points: SKILL.md (scheduled exports, listeners/webhooks).
- Boundary markers: None identified in the flow configurations for isolating data from instructions.
- Capability inventory: SKILL.md (CLI-based flow management, automated data movement, and JavaScript hook execution).
- Sanitization: No explicit sanitization or validation logic for external record content is described in the provided flow structures.
Audit Metadata