orx-lit-review
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data in the form of paper abstracts, snippets, and full text from alphaXiv, OpenAlex, and bioRxiv. This content is then used by the agent to synthesize scholarly claims, creating a surface for potential indirect prompt injection attacks where malicious instructions could be embedded in research paper metadata or content.
- Ingestion points: Data enters the agent's context through the output of
orx discoverandorx papercommands. - Boundary markers: The instructions do not define explicit delimiters or sanitization steps to distinguish between system instructions and external paper content.
- Capability inventory: The skill utilizes the
orxCLI tool to perform network-based discovery and retrieval of academic papers. - Sanitization: There is no evidence of output sanitization or filtering for the external scholarly content processed by the agent.
- [COMMAND_EXECUTION]: The skill relies on the execution of the
orxcommand-line utility to perform its core functions (discover,paper). While these appear to be domain-specific tools for academic retrieval, they represent an active command execution interface within the skill's operational loop.
Audit Metadata