improvement(mcp): align tool discovery timeouts and retries with MCP protocol standard by waleedlatif1 · Pull Request #5753 · simstudioai/sim

waleedlatif1 · 2026-07-18T00:32:06Z

Summary

Raise the tools/list idle timeout from an aggressive 10s → 30s (derived from the per-server timeout, clamped to the max execution timeout) with a 60s hard cap via maxTotalTimeout, matching the MCP TypeScript SDK's DEFAULT_REQUEST_TIMEOUT_MSEC (60s). Enable resetTimeoutOnProgress + an onprogress handler so slow-but-alive servers aren't spuriously failed with McpError -32001.
Retry read-only tools/list on transient transport errors (timeout, connection-closed, 429/5xx, session 404/400, network) with jittered backoff. tools/call stays conservative (session-error retry only) since it isn't idempotent. The SDK doesn't retry POST requests — only SSE streams — so the app owns this.
Wire client.onerror so out-of-band transport failures (SSE disconnects, exhausted reconnects) are observed instead of silently dropped, per the SDK's Protocol.onerror contract.
UX: map timeouts to a clear user-facing message, and surface refresh/remove failures through the standard emcn toast instead of swallowing them.

Context

A workspace MCP server whose tools/list intermittently took >10s was hard-failing discovery (-32001) even though connect/OAuth succeeded — surfacing as a stuck "Connecting…" / "Failed to discover MCP tools" with the tool count dropping from 40 → 0. The 10s cap was ~6× tighter than the SDK default and below every documented ecosystem norm; this loosens it toward the protocol standard and adds the resilience the SDK expects apps to implement themselves.

Type of Change

Bug fix
Improvement

Testing

apps/sim/lib/mcp unit tests (client/service/utils) + MCP route/component/hook tests pass (116 tests).
Added tests: tools/list timeout retries then succeeds; timeout maps to the user-facing lastError; listTools is bounded with idle timeout + hard cap + progress reset.
tsc --noEmit clean; biome clean.

Checklist

Code follows project style guidelines
Self-reviewed my changes
Tests added/updated and passing
No new warnings introduced
I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

…protocol standard - Raise tools/list idle timeout 10s -> 30s (derived from per-server config, clamped to max execution timeout) with a 60s hard cap via maxTotalTimeout, matching the SDK's DEFAULT_REQUEST_TIMEOUT_MSEC. Enable resetTimeoutOnProgress with an onprogress handler so slow-but-alive servers are not spuriously failed. - Retry read-only tools/list on transient transport errors (timeout, connection-closed, 429/5xx, session 404/400, network) with jittered backoff. tools/call stays conservative (session-error retry only) since it is not idempotent. The MCP SDK does not retry POST requests, so the app owns this. - Wire client.onerror so out-of-band transport failures are observed, not silently dropped. - Map timeouts to a user-facing message and surface refresh/remove failures via the standard emcn toast instead of swallowing them.

vercel · 2026-07-18T00:32:12Z

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Jul 18, 2026 12:46am

cursor · 2026-07-18T00:32:16Z

PR Summary

Medium Risk
Changes core MCP discovery and retry behavior used by refresh and workspace tool loading; scope is bounded to read-only tools/list and error handling, with broad test coverage.

Overview
MCP tool discovery is loosened and hardened so slow-but-healthy servers stop failing spuriously and transient outages recover without user action.

tools/list now uses a 30s idle timeout (from per-server config, clamped) and a 60s hard cap, with resetTimeoutOnProgress and progress logging. Discovery retries transient failures (timeouts, connection closed, 429/5xx, stale session 404/400, network blips) with jittered backoff; persisted lastError maps timeouts to a clear user-facing message. The MCP client logs transport errors via client.onerror.

The refresh API wraps post-discovery workflow schema sync in try/catch so a sync failure still returns 200 with connected status and zero workflows updated. MCP settings shows toasts on refresh/remove failures.

^{Reviewed by Cursor Bugbot for commit 561a32f. Configure here.}

greptile-apps · 2026-07-18T00:34:21Z

Greptile Summary

This PR makes MCP tool discovery more resilient and improves failure reporting. The main changes are:

Adds bounded idle and total timeouts for tools/list.
Retries transient discovery failures with jittered backoff.
Records transport errors reported outside normal requests.
Maps discovery timeouts to clearer user-facing messages.
Shows refresh and removal failures through standard toasts.
Keeps successful refreshes from failing when workflow synchronization fails.

Confidence Score: 5/5

This looks safe to merge.

The updated timeout calculation keeps each discovery request within the intended limit.
The new test covers oversized server timeout settings.
No blocking issue remains in the changed fix path.

Important Files Changed

Filename	Overview
apps/sim/lib/mcp/client.ts	Adds transport error logging and applies bounded idle and total timeouts to tool discovery.
apps/sim/lib/mcp/client.test.ts	Covers discovery timeout options and oversized configured timeout values.
apps/sim/lib/mcp/service.ts	Adds transient discovery error classification, retries, backoff, and clearer timeout messages.
apps/sim/lib/mcp/service.test.ts	Covers successful retry behavior and persisted timeout messages.
apps/sim/lib/mcp/utils.ts	Defines separate idle and total timeout constants for tool discovery.
apps/sim/app/api/mcp/servers/[id]/refresh/route.ts	Prevents secondary workflow synchronization failures from turning a successful refresh into an error response.
apps/sim/app/api/mcp/servers/[id]/refresh/route.test.ts	Covers successful refresh responses when workflow synchronization fails.
apps/sim/app/workspace/[workspaceId]/settings/components/mcp/mcp.tsx	Displays standard error toasts when server refresh or removal fails.

_{Reviews (2): Last reviewed commit: "fix(mcp): keep tools/list absolute timeo..." | Re-trigger Greptile}

…nc throws Discovery already persists status and caches tools; a syncToolSchemasToWorkflows failure was escaping the refactored try/catch and returning 500 despite the refresh having succeeded. Guard the secondary sync so it degrades to zero workflows updated instead.

A configured per-server timeout above 60s was expanding maxTotalTimeout past the intended absolute discovery ceiling. Clamp the idle timeout to the ceiling too so tools/list can never hang the UI beyond it; connect() and callTool() still honor the full configured/execution timeout.

waleedlatif1 · 2026-07-18T00:47:06Z

@greptile

waleedlatif1 · 2026-07-18T00:47:07Z

@cursor review

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 561a32f. Configure here.}

greptile-apps Bot reviewed Jul 18, 2026

View reviewed changes

Comment thread apps/sim/lib/mcp/client.ts

chore(mcp): trim comments to match codebase density

0c05adf

vercel Bot temporarily deployed to Preview July 18, 2026 00:36 Inactive

vercel Bot temporarily deployed to Preview July 18, 2026 00:41 Inactive

waleedlatif1 mentioned this pull request Jul 18, 2026

v0.7.39: mcp oauth hardening, chat perf improvements #5752

Merged

vercel Bot temporarily deployed to Preview July 18, 2026 00:46 Inactive

cursor Bot reviewed Jul 18, 2026

View reviewed changes

waleedlatif1 merged commit c739dd6 into staging Jul 18, 2026
20 checks passed

waleedlatif1 deleted the mcp-discovery-resilience branch July 18, 2026 00:50

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

improvement(mcp): align tool discovery timeouts and retries with MCP protocol standard#5753

improvement(mcp): align tool discovery timeouts and retries with MCP protocol standard#5753
waleedlatif1 merged 4 commits into
stagingfrom
mcp-discovery-resilience

waleedlatif1 commented Jul 18, 2026

Uh oh!

vercel Bot commented Jul 18, 2026 •

edited

Loading

Uh oh!

cursor Bot commented Jul 18, 2026 •

edited

Loading

Uh oh!

greptile-apps Bot commented Jul 18, 2026 •

edited

Loading

Uh oh!

Uh oh!

waleedlatif1 commented Jul 18, 2026

Uh oh!

waleedlatif1 commented Jul 18, 2026

Uh oh!

cursor Bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

waleedlatif1 commented Jul 18, 2026

Summary

Context

Type of Change

Testing

Checklist

Uh oh!

vercel Bot commented Jul 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cursor Bot commented Jul 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Summary

Uh oh!

greptile-apps Bot commented Jul 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Uh oh!

Uh oh!

waleedlatif1 commented Jul 18, 2026

Uh oh!

waleedlatif1 commented Jul 18, 2026

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

vercel Bot commented Jul 18, 2026 •

edited

Loading

cursor Bot commented Jul 18, 2026 •

edited

Loading

greptile-apps Bot commented Jul 18, 2026 •

edited

Loading