hybrid-search-implementation
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it is instructed to analyze untrusted data from the local repository to plan and implement changes.
- Ingestion points: Workflow step 1 in
SKILL.mddirects the agent to "Inspect the available files, runtime, versions, inputs, and existing conventions." - Boundary markers: There are no specific instructions to treat content from these files as untrusted or to use delimiters to prevent instruction leakage.
- Capability inventory: The skill has the ability to modify local files, run tests, and execute external API calls through the
sandbase_runtool. - Sanitization: The instructions do not define methods for sanitizing or validating the content of the files read before they are processed by the LLM.
Audit Metadata