Agent-driven MCP testing
Runs an MCP server through Claude Code, Codex, OpenCode, or Pi to test the tool calls those agents actually make.
Loading page…
Coding / product dossier
Tests MCP servers through real coding agents and asserts on their tool calls in pytest.
Product brief
Your MCP unit tests pass, but did the agent call the tool? M3 runs your server through real Claude Code, Codex, OpenCode or Pi and asserts on the calls they actually made, in plain pytest. Direct tests need no API key. m3 ui shows every trace and tool call; m3 ci test --upload gates PRs and keeps runs on app.m3.sineframe.com. Open source, Apache-2.0.
Why we selected it
Tests what coding agents actually do with MCP tools rather than stopping at server unit tests. Open-source pytest integration, traces, and PR gates make it actionable for builders; the news about weak task-completion
Product preview saved with our daily selection
Capability scan
Only capabilities supported by the product information we collected are listed here.
Runs an MCP server through Claude Code, Codex, OpenCode, or Pi to test the tool calls those agents actually make.
Supports assertions on actual agent tool calls in plain pytest.
Provides direct tests that do not require an API key.
The m3 ui command displays traces and tool calls.
The m3 ci test --upload command gates pull requests and keeps runs on app.m3.sineframe.com.
Best-fit use cases
FAQ
M3 runs the server through supported coding agents and lets you assert on the calls they actually make, rather than stopping at server unit tests.
The description names Claude Code, Codex, OpenCode, and Pi.
Direct tests require no API key. The description does not extend that statement to agent-driven tests.
Use m3 ui to view traces and tool calls. The m3 ci test --upload command gates pull requests and keeps runs on app.m3.sineframe.com.
Yes. SineFrame M3 is described as open source under the Apache-2.0 license.