build-cypher-query
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [SAFE]: The skill provides structured instructions for interacting with the SubImage graph database using platform-specific tools like
subimageRunCypherandsubimageGetNodesSchema. These activities are consistent with the skill's primary purpose of graph analysis and do not involve unauthorized data access or exfiltration to external domains. - [COMMAND_EXECUTION]: The skill generates and executes Cypher commands. To mitigate risk, it enforces strict read-only rules, explicitly forbidding writing operations (e.g., CREATE, MERGE, DELETE) and unsafe procedures. It also requires the inclusion of
LIMITclauses to prevent resource exhaustion and denial-of-service scenarios. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data to build database queries. 1. Ingestion points: user questions and schema discovery tool outputs. 2. Boundary markers: not explicitly defined for prompt interpolation. 3. Capability inventory: database query execution via
subimageRunCypher. 4. Sanitization: enforcement of read-only keywords and mandatory result limits. The detailed workflow requiring grounding in validated schema before execution further reduces the risk of malicious injection. - [DYNAMIC_EXECUTION]: The skill dynamically assembles Cypher query strings at runtime. This is handled as a low-risk activity within this context because the execution is restricted to a specialized database language with platform-enforced read-only permissions and defined safety constraints.
Audit Metadata