cloudbase-agent-python
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides a framework for building agents that ingest and process user messages, which is a standard surface for indirect prompt injection attacks.
- Ingestion points: Data enters via the
RunAgentInputmodel defined inreferences/server.mdandadapter-development.md. - Boundary markers: The protocol uses structured message objects (role/content) to separate user input from system instructions.
- Capability inventory: The SDK supports network operations via adapters, tool execution in
references/tools.md, and file system access. - Sanitization: Explicit sanitization of user content within the message stream is not implemented in the provided SDK examples.
- [DYNAMIC_EXECUTION]: The
references/storage.mddocumentation listsPickleSerializeras a supported serialization method. Usingpicklefor deserializing data from a shared or external storage backend (like Redis or PostgreSQL) can lead to arbitrary code execution if the stored data is manipulated by an attacker.
Audit Metadata