create-adapter
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes how to build adapters that ingest conversation data from external sources like Cursor databases and Warp SQLite files. This creates an indirect prompt injection surface where instructions hidden in the source data could be processed by the agent.\n
- Ingestion points: SQLite databases (store.db, warp.sqlite) and file-based logs mentioned in SKILL.md and the reference files.\n
- Boundary markers: While the skill suggests "Message Content Cleaning" and "stripping prefixes," it does not explicitly mandate robust boundary delimiters for the imported content.\n
- Capability inventory: The Adapter interface requires methods for reading sessions and messages from the local filesystem and databases.\n
- Sanitization: The documentation recommends defensive metadata parsing and stripping channel/origin tags from message content.
Audit Metadata