Validate all wire messages against the per-version spec JSON schema (#399)

* Vendor schema/{version}/schema.json alongside schema.ts

sync-schema now copies the spec repo's JSON Schema for each version into
src/spec-types/{version}.schema.json (same pinned SHA recorded in SOURCE:
modelcontextprotocol@f817239f4d6b1efff2c4dfc2f7af85c985d73076). These are
the inputs for runtime wire-message validation. Vendored verbatim, so they
join schema.ts in .prettierignore.

Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JHNrpU3Vr6Q1rPZiji7zLY

* Validate all wire messages against the per-version spec JSON schema

Every JSON-RPC message the harness sends or receives is now validated
(ajv) against the vendored schema.json for the run's spec version, at the
wire choke points: sendStatelessRequest (outbound request, JSON response,
SSE events), connectStateful (request/result/notifications), and both mock
servers (inbound client requests, outbound responses/handler results).

Beyond the JSONRPCMessage envelope, messages are validated against the
most specific definition the schema has: typed requests/notifications by
method const, typed error responses by error.code const (e.g. -32021 ->
MissingRequiredClientCapabilityError, the #376 hallucination), and typed
results via the XxxRequest -> XxxResult pair, with SEP-2322
InputRequiredResult discriminated on resultType.

Violations are recorded per scenario and surfaced by the runners as a
synthetic wire-schema-valid check (FAILURE when the implementation under
test sent an invalid message) plus a wire-schema-harness-error check when
the invalid message was harness-authored. A vitest-wide afterEach guard
fails any self-test that recorded violations, so hallucinated fixtures
can't ship (#376, #271). Opt-outs are explicit and greppable: per-call
skipValidation for deliberately malformed traffic, takeWireViolations()
drains in tests that exercise deliberately broken fixtures.

Offenders the new validation caught, now fixed:
- everything-server stateless path omitted resultType (and the caching
  hints on server/discover) required by the draft schema
- sep-2164-empty-contents fixture omitted resultType/ttlMs/cacheScope
  (its intended violation, empty contents, is schema-valid prose-only)
- connection/mock-server unit-test fixtures used schema-invalid stubs
  (tools/call without name, ListToolsResult without required members,
  Tool without inputSchema, ProgressNotification without progressToken,
  ElicitRequest without message/requestedSchema)
- http-standard-headers test mocks returned bare { tools: [] } results
- input-required-result-validate-input deliberately sends malformed
  inputResponses; now marked skipValidation

Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JHNrpU3Vr6Q1rPZiji7zLY

* Validate the stateful wire at the SDK transport, close review-panel gaps

Address the review-panel findings on the wire-schema validation PR:

- Stateful wire: hook the SDK transport inside connectToServer, wrapping
  send/onmessage so every raw wire message in both directions is
  validated for the scenario's spec version -- including the initialize
  handshake, server->client requests (elicitation/sampling), and the
  harness's responses to them. Request ids are correlated per direction
  so responses validate against their typed result definition (e.g. an
  outbound ElicitResult). The reconstructed stand-ins in
  connectStateful (fake id: 0 envelopes, Zod-reparsed results) are
  removed; the hook sees the real bytes. The four scenarios that call
  connectToServer directly (elicitation-defaults, elicitation-enums,
  tools, lifecycle) now pass ctx.specVersion and get validation for
  free; the runner surfaces their violations through the existing
  recorder. Known remaining gap (documented in the module doc): the
  client-auth scenarios' bespoke express mock is not instrumented.

- skipValidation now skips only the request direction; the
  implementation's response to a deliberately malformed request is
  still validated, with a narrow carve-out tolerating the JSON-RPC 2.0
  id: null an error response must carry when the request could not be
  processed (the MCP schema's RequestId forbids null). Covered by new
  unit tests pinning both the still-validated response and the
  narrowness of the carve-out.

- negative-mrtr.test.ts: the file-wide drain now asserts every recorded
  violation is implementation-origin (the deliberately broken fixture)
  and throws on any harness-origin violation.

- everything-server: hoist the cacheable-methods set to module scope
  and replace the per-request res.json monkey-patch with an explicit
  sendStatelessJson(res, method, payload) helper applied at each
  dispatch site.

- Pin the extracted dispatch maps (methodDefs/errorDefs, e.g.
  -32021 -> MissingRequiredClientCapabilityError under the draft) per
  spec version in wire-schema.test.ts, so a schema sync that breaks the
  const extraction fails loudly instead of degrading to envelope-only
  validation. Also note the batch branch's requestMethod forwarding
  limitation.

Offender the transport hook flushed out: the everything-server's
test_elicitation_sep1034_defaults tool sent a number-typed elicitation
field with default: 95.5, which the 2025-11-25 schema forbids
(NumberSchema.default is typed integer there; widened to number only in
the draft). The fixture and the scenario's expected default are now the
integer-valued 95, with comments explaining why.

Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JHNrpU3Vr6Q1rPZiji7zLY

* Condense comments to three lines max

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JHNrpU3Vr6Q1rPZiji7zLY

* Re-vendor spec schemas at pinned SHA 71e30695

Regenerate src/spec-types/*.schema.json via npm run sync-schema at the
SHA already recorded in SOURCE. The previously vendored draft schema
predated spec PR #3002, so it still required DiscoverResult.serverInfo
at the top level; the mock server now correctly reports serverInfo via
result _meta, which the stale schema rejected.

* Scope the wire-violation recorder per scenario for parallel suite runs

Client --suite runs execute scenarios concurrently over what was a
module-global violation recorder, so scenarios could steal or wipe each
other's recorded violations. Give each suite scenario its own recorder
via AsyncLocalStorage; code outside a scope (single-scenario runs, the
sequential server suite, vitest hooks) falls back to the global
recorder. Also condense a four-line comment in connection/stateful.ts
to three lines.

* Patch the 2025-11-25 NumberSchema generator bug as a load-time erratum

The released 2025-11-25 schema.json types NumberSchema
minimum/maximum/default as integer, contradicting its own schema.ts
(default?: number) — a typescript-json-schema artifact, fixed for draft
in modelcontextprotocol#2710 and proposed for 2025-11-25 in
modelcontextprotocol#3139. Patch the schema at load time so SDK
fixtures with fractional number defaults (all tier-1 SDKs send 95.5)
are not flagged; restore the sep-1034 scenario and example-server
fixture to 95.5. Delete the erratum when #3139 lands and the schemas
are re-vendored.

Also scope the sequential server-suite loop with withWireRecorder so a
connection leaked by a failed scenario cannot attribute late traffic to
the next scenario.

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Paul Carleton <paulc@anthropic.com>
This commit is contained in:
claude[bot] 2026-07-27 15:58:43 +01:00 • committed by GitHub
commit 89c2c0452e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 14136 additions and 129 deletions

View file

@ -1235,6 +1235,36 @@ const LEGACY_SESSION_PROTOCOL_VERSIONS = [
'2025-11-25'
];
// Stateless (draft) operations whose results MUST carry the SEP-2549 caching
// hints (`ttlMs`, `cacheScope`).
const STATELESS_CACHEABLE_METHODS: ReadonlySet<string> = new Set([
'server/discover',
'tools/list',
'prompts/list',
'resources/list',
'resources/templates/list',
'resources/read'
]);
/** Send a stateless (draft) JSON-RPC response. Draft results MUST carry `resultType`
* and cacheable operations the SEP-2549 caching hints; stamp any the dispatch site did
* not set so every stateless result is draft-schema-valid. Errors pass through untouched. */
function sendStatelessJson(
res: import('express').Response,
method: string,
payload: { result?: Record<string, unknown>; [key: string]: unknown }
): import('express').Response {
const result = payload.result;
if (result && typeof result === 'object' && !Array.isArray(result)) {
result.resultType ??= 'complete';
if (STATELESS_CACHEABLE_METHODS.has(method)) {
result.ttlMs ??= 0;
result.cacheScope ??= 'private';
}
}
return res.json(payload);
}
// Handle POST requests - stateful mode
app.post('/mcp', async (req, res) => {
const sessionId = req.headers['mcp-session-id'] as string | undefined;
@ -1360,7 +1390,7 @@ app.post('/mcp', async (req, res) => {
}
if (method === 'server/discover') {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1390,7 +1420,7 @@ app.post('/mcp', async (req, res) => {
{ method: 'tools/list', params: {} },
ResultSchema as any
)) as { tools: any[]; [k: string]: unknown };
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1466,7 +1496,7 @@ app.post('/mcp', async (req, res) => {
}
});
} catch (e: any) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
error: { code: e.code ?? -32603, message: e.message, data: e.data }
@ -1484,7 +1514,7 @@ app.post('/mcp', async (req, res) => {
{ method: 'prompts/list', params: {} },
ResultSchema as any
)) as { prompts: any[]; [k: string]: unknown };
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1501,7 +1531,7 @@ app.post('/mcp', async (req, res) => {
}
});
} catch (e: any) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
error: { code: e.code ?? -32603, message: e.message, data: e.data }
@ -1522,7 +1552,7 @@ app.post('/mcp', async (req, res) => {
inputResponses['user_context'],
'context'
);
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1538,7 +1568,7 @@ app.post('/mcp', async (req, res) => {
}
});
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1572,7 +1602,7 @@ app.post('/mcp', async (req, res) => {
{ method: 'resources/list', params: {} },
ResultSchema as any
)) as { resources: any[]; [k: string]: unknown };
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1591,7 +1621,7 @@ app.post('/mcp', async (req, res) => {
}
});
} catch (e: any) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
error: { code: e.code ?? -32603, message: e.message, data: e.data }
@ -1608,7 +1638,7 @@ app.post('/mcp', async (req, res) => {
{ method: 'resources/templates/list', params: {} },
ResultSchema as any
)) as { resourceTemplates: any[]; [k: string]: unknown };
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1618,7 +1648,7 @@ app.post('/mcp', async (req, res) => {
}
});
} catch (e: any) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
error: { code: e.code ?? -32603, message: e.message, data: e.data }
@ -1631,7 +1661,7 @@ app.post('/mcp', async (req, res) => {
if (method === 'resources/read') {
const uri = params.uri as string | undefined;
if (uri === 'test://stateless-static-text') {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1675,7 +1705,7 @@ app.post('/mcp', async (req, res) => {
}
});
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: { content: [{ type: 'text', text: 'Success' }] }
@ -1687,13 +1717,13 @@ app.post('/mcp', async (req, res) => {
if (name === 'test_input_required_result_elicitation') {
if (inputResponses?.['user_name']) {
const userName = getMrtInputText(inputResponses['user_name'], 'name');
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: { content: [{ type: 'text', text: `Hello, ${userName}!` }] }
});
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1722,7 +1752,7 @@ app.post('/mcp', async (req, res) => {
unknown
>;
const content = sample.content as Record<string, unknown> | undefined;
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1735,7 +1765,7 @@ app.post('/mcp', async (req, res) => {
}
});
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1770,7 +1800,7 @@ app.post('/mcp', async (req, res) => {
const roots = Array.isArray(rootsResult.roots)
? rootsResult.roots
: [];
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1778,7 +1808,7 @@ app.post('/mcp', async (req, res) => {
}
});
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1796,7 +1826,7 @@ app.post('/mcp', async (req, res) => {
const ok = (inputResponses['confirm'] as Record<string, unknown>)
?.content as Record<string, unknown> | undefined;
if (state.kind === 'request-state' && ok?.ok === true) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1807,7 +1837,7 @@ app.post('/mcp', async (req, res) => {
});
}
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1860,7 +1890,7 @@ app.post('/mcp', async (req, res) => {
const roots = Array.isArray(rootsResult.roots)
? rootsResult.roots
: [];
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1874,7 +1904,7 @@ app.post('/mcp', async (req, res) => {
});
}
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1915,7 +1945,7 @@ app.post('/mcp', async (req, res) => {
if (name === 'test_input_required_result_multi_round') {
if (!requestState) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1940,7 +1970,7 @@ app.post('/mcp', async (req, res) => {
const state = JSON.parse(requestState) as Record<string, unknown>;
if (state.round === 1 && inputResponses?.['step1']) {
const userName = getMrtInputText(inputResponses['step1'], 'name');
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1970,7 +2000,7 @@ app.post('/mcp', async (req, res) => {
const userName =
typeof state.name === 'string' ? state.name : 'friend';
const color = getMrtInputText(inputResponses['step2'], 'color');
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -1984,7 +2014,7 @@ app.post('/mcp', async (req, res) => {
});
}
// Fallback: restart
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -2011,7 +2041,7 @@ app.post('/mcp', async (req, res) => {
if (requestState) {
const verified = verifyMrtState(requestState);
if (!verified) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
error: {
@ -2021,7 +2051,7 @@ app.post('/mcp', async (req, res) => {
});
}
if (verified.kind === 'tamper-test' && inputResponses?.['confirm']) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -2032,7 +2062,7 @@ app.post('/mcp', async (req, res) => {
});
}
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -2093,7 +2123,7 @@ app.post('/mcp', async (req, res) => {
}
if (inputResponses && Object.keys(inputResponses).length > 0) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -2107,7 +2137,7 @@ app.post('/mcp', async (req, res) => {
});
}
if (Object.keys(inputRequests).length === 0) {
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -2117,7 +2147,7 @@ app.post('/mcp', async (req, res) => {
}
});
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: {
@ -2197,7 +2227,7 @@ app.post('/mcp', async (req, res) => {
} else {
notifyListenStreams('prompts', 'notifications/prompts/list_changed');
}
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
result: { content: [{ type: 'text', text: 'Mutation triggered' }] }
@ -2255,14 +2285,14 @@ app.post('/mcp', async (req, res) => {
{ method, params },
ResultSchema as any
);
return res.json({ jsonrpc: '2.0', id, result });
return sendStatelessJson(res, method, { jsonrpc: '2.0', id, result });
} catch (e: any) {
// SEP-2164: unknown resources get -32602 with the requested uri in
// data; the SDK's McpError does not populate data itself.
const data =
e.data ??
(method === 'resources/read' ? { uri: params.uri } : undefined);
return res.json({
return sendStatelessJson(res, method, {
jsonrpc: '2.0',
id,
error: { code: e.code ?? -32603, message: e.message, data }

View file

@ -7,6 +7,9 @@
* contents array for any resources/read request, violating the SEP-2164 MUST
* NOT. The sep-2164-resource-not-found scenario should emit FAILURE for
* sep-2164-no-empty-contents against this server.
*
* Responses are otherwise schema-valid (resultType and SEP-2549 caching hints present);
* an empty contents array is schema-valid too, forbidden only by the SEP-2164 prose.
*/
import express from 'express';
@ -25,16 +28,37 @@ app.post('/mcp', (req, res) => {
jsonrpc: '2.0',
id,
result: {
resultType: 'complete',
ttlMs: 0,
cacheScope: 'private',
supportedVersions: ['2026-07-28'],
capabilities: { resources: {} },
serverInfo: { name: 'sep-2164-empty-contents', version: '1.0.0' }
}
});
case 'resources/list':
return res.json({ jsonrpc: '2.0', id, result: { resources: [] } });
return res.json({
jsonrpc: '2.0',
id,
result: {
resultType: 'complete',
ttlMs: 0,
cacheScope: 'private',
resources: []
}
});
case 'resources/read':
// Deliberately return an empty contents array instead of an error.
return res.json({ jsonrpc: '2.0', id, result: { contents: [] } });
return res.json({
jsonrpc: '2.0',
id,
result: {
resultType: 'complete',
ttlMs: 0,
cacheScope: 'private',
contents: []
}
});
default:
return res.status(404).json({
jsonrpc: '2.0',