Oracle Agent Spec Expert
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill follows security best practices by explicitly advising against hardcoding credentials and recommending the use of environment variables (e.g.,
{env.API_KEY}). The design patterns provided are conceptual and educational in nature. - [INDIRECT_PROMPT_INJECTION]: The specification describes an architecture where agents ingest data from external sources (APIs, databases) and interpolate it into LLM prompts. This creates a potential surface for indirect prompt injection if user-controlled data is not properly handled by the underlying execution runtime.
- Ingestion points: Data entering through
APINodeparameters and{input.location}or{gather.output}interpolation inLLMNodetemplates inSKILL.md. - Boundary markers: Not explicitly demonstrated in the conceptual YAML/JSON examples.
- Capability inventory: The specification supports text generation, external API requests, and database queries.
- Sanitization: The skill addresses this surface in the "Best Practices" section, recommending that developers define explicit input/output schemas and validation to mitigate injection risks.
- [EXTERNAL_DOWNLOADS]: The skill references resources from a trusted organization's GitHub repository and official documentation sites. It also mentions the installation of a standard utility package from a public registry (PyPI).
Audit Metadata