Coding agents
The plugin installs from one marketplace and runs the same flows on each coding agent. Only the way you install it and type a command differs. Claude Code, GitHub Copilot CLI and Codex CLI are all supported. Cursor support is coming soon.Claude Code
The primary target. Commands start with
/, for example /simulate.Codex CLI
Commands start with
$, for example $simulate.Copilot CLI
Commands carry the plugin name, for example
/plurai-simulation:simulate.Agent frameworks
The harness (simulate.py) drives your agent without changing its source. Python agents run in-process;
Node agents run in their own runtime through a bridge.
Python agent frameworks
LangGraph, Claude Agent SDK, Google ADK, CrewAI, OpenAI Agents SDK, LlamaIndex, Strands, LangChain, or no framework at all.
Node agents
Mastra, Vercel AI SDK, LangGraph.js, OpenAI Agents JS or a custom loop, through the Node bridge.
Observability platforms
Plurai Simulation works natively with LangSmith, Langfuse, Arize and Braintrust. Ground the mock environment, scenarios and evals in real production behavior, replay real conversations from the turn where they broke, and record Runs back to the platform your team already uses.Observability platforms
Analyze traces from LangSmith and Langfuse, replay conversations from LangSmith and Arize, and record
Runs to all four.
Python SDK
Use the
plurai SDK directly, for CI or for a harness you write yourself.