ci: use conformance composite action instead of npx scripts

Replace manual npx conformance invocations with the composite GitHub
Action from modelcontextprotocol/conformance. This:

- Removes continue-on-error (replaced by expected-failures baseline)
- Adds conformance-baseline.yml for tracking known failures
- Builds conformance tool from source at the referenced commit
- Passes node-version: 24 to match SDK requirements

Author: maxisbey

.github/workflows/conformance.yml

@@ -1,51 +1,65 @@
 name: Conformance Tests
 
 on:
-    push:
-        branches: [main]
-    pull_request:
-    workflow_dispatch:
+  push:
+    branches: [main]
+  pull_request:
+  workflow_dispatch:
 
 concurrency:
-    group: conformance-${{ github.ref }}
-    cancel-in-progress: true
+  group: conformance-${{ github.ref }}
+  cancel-in-progress: true
 
 permissions:
-    contents: read
+  contents: read
 
 jobs:
-    client-conformance:
-        runs-on: ubuntu-latest
-        continue-on-error: true
-        steps:
-            - uses: actions/checkout@v4
-            - name: Install pnpm
-              uses: pnpm/action-setup@v4
-              with:
-                  run_install: false
-            - uses: actions/setup-node@v4
-              with:
-                  node-version: 24
-                  cache: pnpm
-                  cache-dependency-path: pnpm-lock.yaml
-            - run: pnpm install
-            - run: pnpm run build:all
-            - run: pnpm run test:conformance:client:all
+  client-conformance:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          run_install: false
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+          cache-dependency-path: pnpm-lock.yaml
+      - run: pnpm install
+      - run: pnpm run build:all
+      - uses: modelcontextprotocol/conformance@4a42c25058c4636aca1ae7f6547ee30805de36ee
+        with:
+          mode: client
+          command: "npx tsx src/conformance/everything-client.ts"
+          suite: all
+          expected-failures: ./conformance-baseline.yml
+          node-version: "24"
 
-    server-conformance:
-        runs-on: ubuntu-latest
-        continue-on-error: true
-        steps:
-            - uses: actions/checkout@v4
-            - name: Install pnpm
-              uses: pnpm/action-setup@v4
-              with:
-                  run_install: false
-            - uses: actions/setup-node@v4
-              with:
-                  node-version: 24
-                  cache: pnpm
-                  cache-dependency-path: pnpm-lock.yaml
-            - run: pnpm install
-            - run: pnpm run build:all
-            - run: pnpm run test:conformance:server
+  server-conformance:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          run_install: false
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+          cache-dependency-path: pnpm-lock.yaml
+      - run: pnpm install
+      - run: pnpm run build:all
+      - name: Start everything-server
+        run: |
+          npx tsx src/conformance/everything-server.ts &
+          timeout 15 bash -c 'until curl -s http://localhost:3000/mcp > /dev/null 2>&1; do sleep 0.5; done'
+          echo "Server ready"
+      - uses: modelcontextprotocol/conformance@4a42c25058c4636aca1ae7f6547ee30805de36ee
+        with:
+          mode: server
+          url: http://localhost:3000/mcp
+          expected-failures: ./conformance-baseline.yml
+          node-version: "24"

conformance-baseline.yml

@@ -0,0 +1,13 @@
+# MCP Conformance Test Expected Failures
+# List scenario names that are known to fail in this SDK.
+# Format:
+#   server:
+#     - scenario-name-1
+#     - scenario-name-2
+#   client:
+#     - scenario-name-3
+#
+# The conformance runner will:
+#   - Exit 0 if only expected failures occur
+#   - Exit 1 if unexpected failures occur (regression)
+#   - Exit 1 if expected failures now pass (stale baseline, update this file)