oar-authoring
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to ingest and analyze external configuration files (config.json, user.json, and _conditions.txt), which serves as a surface for indirect prompt injection if the files contain malicious natural language instructions intended to influence the agent.\n
- Ingestion points: The agent is directed to read submod and mod-level configuration files from game-related directories (e.g., meshes/actors/.../OpenAnimationReplacer/).\n
- Boundary markers: There are no prescribed delimiters or explicit warnings to ignore natural language content within the ingested configuration data.\n
- Capability inventory: The skill leverages file operations such as Read, Glob, Grep, and Write, and utilizes domain-specific tools like housecarl_read_record and housecarl_cross_plugin_query to interact with the system.\n
- Sanitization: No procedures are specified for validating or sanitizing the content of the configuration files before they are processed by the agent.
Audit Metadata