investigate-container
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied identifiers such as image URIs, digests, and cluster names to query a backend graph, creating a potential surface for indirect prompt injection if malicious data is provided. Ingestion points: SKILL.md defines inputs sourced from user prompts. Boundary markers: Absent; user data is interpolated into query strings without formal delimiters. Capability inventory: The skill uses subimageRunCypher (documented in references/cypher-templates.md) for database operations. Sanitization: Present; references/cypher-templates.md instructs the agent to manually escape backslashes and single quotes before substitution.
- [DYNAMIC_EXECUTION]: The skill generates and executes database queries (Cypher) at runtime based on user-supplied variables. Evidence: references/cypher-templates.md provides templates with placeholders that the agent is instructed to populate with user data and execute via the subimageRunCypher tool.
Audit Metadata