Declarative scenarios
Describe actors, steps, inputs, and assertions in YAML. No pytest code required for standard conformance coverage.
mcp-conformance is a scenario-driven test runner that verifies MCP implementations against the protocol specification. It runs declarative YAML scenarios against configurable test partners and produces structured pass/fail reports suitable for CI gates and enterprise audit evidence.
Describe actors, steps, inputs, and assertions in YAML. No pytest code required for standard conformance coverage.
Every step captures the full HTTP request/response, JSON-RPC message, redirect chain, and elapsed time. Failures come with evidence.
Default partner is mcp-auth-test-server. Third-party adapters can target any MCP server. Plug in and run the same scenario catalog.
# Clone and install
git clone https://github.com/rmax-ai/mcp-conformance.git
cd mcp-conformance
uv sync --dev
# Run unit tests
uv run pytest tests/unit/ -v
# Run the full scenario suite (requires mcp-auth-test-server)
mcp-conformance run MCP implementations are proliferating across clients, SDKs, gateways, and hosted servers. There is no shared mechanism to verify whether a client handles the full authorization path, whether a server returns spec-shaped JSON-RPC errors, or whether an OAuth server validates PKCE correctly.
mcp-conformance solves this with:
scenarios/ (YAML) → Scenario Runner → Partner Adapter → Test Server
↓
Assertion Library
↓
JSON Report MIT