flexport-migration-deep-dive

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines processes for importing data from legacy logistics systems and databases which constitutes an attack surface for indirect prompt injection.
  • Ingestion points: The migrateProducts, validateRoute, and decommissionLegacy functions in SKILL.md ingest external records.
  • Boundary markers: None are defined in the code templates to separate data from instructions.
  • Capability inventory: The skill utilizes Read, Write, Edit, Bash, and fetch tools.
  • Sanitization: Example code snippets lack input validation or escaping for the ingested data.
  • [DATA_EXFILTRATION]: The skill includes code templates that perform network requests to api.flexport.com. This activity is expected for the skill's purpose of integrating with the Flexport API and targets a well-known service domain.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 08:37 AM
Security Audit — agent-trust-hub — flexport-migration-deep-dive