- add .idea/ for Jetbrains IDEs
* in everything.ts
- remove import of SetLevelRequestSchema
- remove logLevel var
- add sessionId var
- in startNotificationIntervals function
- add optional sid argument
- set sessionId to sid
- define messages to be sent, adding sessionId if present
- remove setRequestHandler call for SetLevelRequestSchema
- replace server.notification calls that sent "notifications/message" objects with calls to server.sendLoggingMessage, passing just the parameters and sessionId.
* In package.json & package-lock.json
- bump TS SDK version to 1.17.5
* In sse.ts, pass transport.sessionId to startNotificationIntervals call
* In stdio.ts
- destructure startNotificationIntervals from createServer call
- implement custom logging request handler and server.sendLoggingMessage implementation, as a
workaround for the fact that the SDK's automatic log level handling currently only tracks requested log level by session id. This will be fixed in a followup PR for the SDK
- call the startNotificationIntervals function after connecting the transport to the server
* In streamableHttp.ts
- destructure startNotificationIntervals from createServer call
- call startNotificationIntervals passing the transport.sessionId after connecting the transport to the server
Fixes issue #474 - documentation mismatch where temperature was documented as (number) but TypeScript SDK only accepts string arguments via z.record(z.string())
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: Ola Hungerford <olaservo@users.noreply.github.com>
* In everything.ts
- remove inappropriate elicitation entry from server capabilities (this is a client capability)
- When creating tool list, only add `ToolName.ELICITATION` definition to tools array if `clientCapabilities` includes `elicitation`
* In package.json / package-lock.json
- bump @modelcontextprotocol/sdk to "^1.17.4", adding `elicitation` to `ClientCapabilities` type
Add Mandoline AI evaluation framework MCP server to the community servers section.
Enables AI assistants to reflect on and continuously improve their performance.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add clientSupportsRoots tracking variable
- Set clientSupportsRoots during initialization based on client capabilities
- Update listRoots tool to provide clearer messaging:
- Specific message when client doesn't support roots protocol
- Different message when client supports roots but none are configured
- Improves user experience by clearly explaining the different scenarios
Addresses feedback from @olaservo in PR review
