You'll notice
Useful3 Signal3
Simultaneous MCP re-auth attempts now collapse into one retry instead of racing each other.
What's wrong with this entry?
Anonymous. No account, no email.
What
When an MCP server needs reauthenticating, concurrent attempts now collapse into a single in-flight retry that reports itself as started, joined or stale_refused, alongside new per-connection bookkeeping for superseded connection attempts and background refreshes. Two accompanying hooks exist only for tests and cannot be used here.
Details
- The setter that would keep a stale reauth entry around throws on any truthy value, so the behaviour it guards is unreachable in this build.
- The hook for observing reauth decisions has no caller in the bundle and is read with optional chaining at two places, making it a no-op.
Evidence
setHoldStaleReauthEntryForTest is test-only
Strings lifted out of the shipped bundle, so the claim above can be checked against them.