Enables deterministic security testing of AI agents that use tools by serving synthetic MCP environments with poisoned data, fake secrets, and privileged actions. Records agent tool calls and evaluates security invariants (e.g., canary leaks, forbidden access, approval binding) without an LLM judge or real systems.
Enables defenders to deploy a decoy MCP tool server that records and fingerprints how LLM agents probe, escalate, and persist, without exposing real systems.
Enables MCP-capable LLM clients to perform read-only Linux system observation and OS algorithm experiments by exposing typed tools for memory, filesystem, process, and CPU scheduling data with controlled, safe boundaries.
A FastMCP server exposing 22 tools for calendar, to-do, notes, web search, math, scratchpad, task queue, and sandboxed code execution, designed for safe RL training with structured outputs and FastMCP transforms.
Enables LLM-driven tool execution with policy-gated authorization, deterministic verification, and replay for radiographic measurement and SQL repair tasks.
