logos-router-distributed-reasoning
Fail
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Fetches and executes the installation script from Ollama's official domain (https://ollama.com/install.sh) by piping it to the shell.
- [EXTERNAL_DOWNLOADS]: Downloads source code from a third-party GitHub repository (https://github.com/rak7777/mythic-mcp-proxy.git) and installs associated Python dependencies.
- [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection through its ingestion of untrusted data.
- Ingestion points: Untrusted data enters the agent context through prompt strings in the
querymethod and input data withinworkflow.executecalls. - Boundary markers: No delimiters or instructions are used to distinguish between system prompts and external data inputs.
- Capability inventory: The skill manages a distributed node mesh, performs inference, and provides a CLI for mesh administration.
- Sanitization: There is no evidence of content validation or sanitization before data is processed by the reasoning units.
Recommendations
- HIGH: Downloads and executes remote code from: https://ollama.com/install.sh - DO NOT USE without thorough review
Audit Metadata