graph-database-expert
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: Hardcoded default credentials 'root/root' for a local SurrealDB instance are provided in test fixtures within SKILL.md.
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with and processes external graph data, which presents a surface for indirect prompt injection if the retrieved content contains malicious instructions.
- Ingestion points: SurrealQL query results are processed and returned to the agent context as shown in SKILL.md and src/graph/queries.py examples.
- Boundary markers: None explicitly defined in the provided code snippets for agent prompt interpolation.
- Capability inventory: The skill utilizes the surrealdb package for network communication with database servers and encourages the use of pytest for local command execution.
- Sanitization: The skill correctly prioritizes parameterized queries as a critical mitigation against injection into the database layers.
Audit Metadata