Title: SEP-992: Notification Configuration for Tool Call Result · Issue #992 · modelcontextprotocol/modelcontextprotocol · GitHub
Open Graph Title: SEP-992: Notification Configuration for Tool Call Result · Issue #992 · modelcontextprotocol/modelcontextprotocol
X Title: SEP-992: Notification Configuration for Tool Call Result · Issue #992 · modelcontextprotocol/modelcontextprotocol
Description: Preamble Authors: Mathis Joffre Abstract This proposal describes a mechanism for configuring notification routing when tool calls complete. Using this mechanism: Clients can specify where and how to receive notifications about tool call ...
Open Graph Description: Preamble Authors: Mathis Joffre Abstract This proposal describes a mechanism for configuring notification routing when tool calls complete. Using this mechanism: Clients can specify where and how t...
X Description: Preamble Authors: Mathis Joffre Abstract This proposal describes a mechanism for configuring notification routing when tool calls complete. Using this mechanism: Clients can specify where and how t...
Mail addresses
mjoffre@blaxel.ai
Opengraph URL: https://github.com/modelcontextprotocol/modelcontextprotocol/issues/992
X: @github
Domain: github.com
{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"SEP-992: Notification Configuration for Tool Call Result","articleBody":"## Preamble\n\n**Authors:** [Mathis Joffre](mailto:mjoffre@blaxel.ai)\n\n## Abstract\n\nThis proposal describes a mechanism for configuring notification routing when tool calls complete. Using this mechanism:\n\n- Clients can specify where and how to receive notifications about tool call results\n- Servers can deliver notifications through transport-appropriate mechanisms\n- Clients can configure triggers based on tool call outcomes (success, failure, timeout)\n- Notifications work seamlessly with resumable requests when both features are enabled\n\n## Motivation\n\n- **Enabling asynchronous tool result delivery**\n - Currently, clients must maintain active connections to receive tool call results\n - There is no standard way to configure alternative notification delivery mechanisms\n - Long-running tools may complete after clients have disconnected\n - Different transports have different notification capabilities that aren't being utilized\n- **Supporting flexible notification routing**\n - Clients need to route different tool results to different endpoints\n - Some results may require immediate notification while others can be batched\n - Transport-specific notification mechanisms (e.g., Unix signals for STDIO, webhooks for HTTP) need a unified configuration approach\n\n## Specification\n\n### 1. Capability negotiation\n\nServers supporting notification routing **MUST** declare their supported route types, along the fact they support notification, via capability negotiation:\n\n```ts\ninterface ServerCapabilities {\n ...\n notificationConfig?: {\n supportedRouteTypes: string[];\n };\n}\n```\n\n### 2. Client request\n\nClients **MAY** advertise a `notificationConfig` capability. When making a `tools/call` request, clients **MAY** include a `notificationConfig` parameter to request notification routing:\n\n```json\n{\n \"jsonrpc\": \"2.0\",\n \"id\": \"123\",\n \"method\": \"tools/call\",\n \"params\": {\n \"name\": \"data_analysis\",\n \"arguments\": { ... },\n \"notificationConfig\": {\n \"routes\": [\n {\n \"id\": \"primary\",\n \"type\": \"webhook\",\n \"endpoint\": \"https://example.com/notifications\",\n \"triggers\": [\"completed\", \"failed\"]\n }\n ]\n }\n }\n}\n```\n\nThe `routes` array defines prioritized routes. The server selects, for each trigger, the first supported route it can deliver to. Servers **MUST** validate that all requested route types are supported and **MUST** reject the request with an error if unsupported types are used.\n\n### 3. Acknowledgement \n\nServers that support `notificationConfig` **MUST** acknowledge the configuration:\n \n```json\n {\n \"jsonrpc\": \"2.0\",\n \"method\": \"notifications/config/acknowledged\",\n \"params\": {\n \"requestId\": \"123\",\n \"acceptedRoutes\": [\"primary\"]\n }\n }\n``` \n\n### Trigger Types\n\nThe following trigger types are defined:\n\n- `completed` — Tool call completed successfully\n- `failed` — Tool call failed with an error\n- `timeout` — Tool call exceeded time limit\n- `cancelled` — Tool call was cancelled\n- `streaming` — Tool call emitted a streaming response (e.g., via events or chunks)\n\nTrigger types are extensible. Additional types may be introduced in future SEPs.\n\n### Server Behavior\n\nWhen a configured trigger fires, the server:\n\n- Selects the first route (by priority) that it supports and can deliver on\n- Sends the notification using the configured transport and payload format\n- Optionally also delivers the result to the client if still connected\n\nIf no supported route is available for a given trigger, the server **MAY** silently drop the notification or include an error in the final result. It **MUST** do one of the two.\n\n### Notification Delivery Semantics\n\nTransport-specific behavior for each `type` (e.g., webhook, signal) will be defined in separate SEPs.\n\nServers **SHOULD** implement:\n- Retry policies for transient errors\n- Rate limiting to protect endpoints\n- Filtering or masking of sensitive data if necessary\n\n### Example Flow\n\n```mermaid\nsequenceDiagram\n participant Client\n participant Server\n participant Tool\n participant NotificationEndpoint as Notification Endpoint\n\n Note over Client,Server: Client advertises notificationConfig capability\n\n Client-\u003e\u003eServer: tools/call with notificationConfig\n Server-\u003e\u003eTool: Execute tool\n\n alt Tool Success\n Tool-\u003e\u003eServer: Result\n Note over Server: Trigger \"completed\"\n Server-\u003e\u003eNotificationEndpoint: Send notification\n Server-\u003e\u003eClient: Tool result (if connected)\n else Tool Failure\n Tool-\u003e\u003eServer: Error\n Note over Server: Trigger \"failed\"\n Server-\u003e\u003eNotificationEndpoint: Send notification\n Server-\u003e\u003eClient: Error (if connected)\n else Tool Timeout\n Note over Server: Trigger \"timeout\"\n Server-\u003e\u003eNotificationEndpoint: Send notification\n Server-\u003e\u003eClient: Timeout (if connected)\n end\n\n Note over Client,NotificationEndpoint: Notifications are sent even if client is disconnected\n```\n\n## Rationale\n\nThe above specification addresses the issues outlined in the Motivation:\n\n- The `notificationConfig` parameter provides a standard way to configure notifications across all transports\n- The route-based system allows flexible notification delivery without prescribing specific mechanisms\n- Trigger types provide a consistent way to specify when notifications should be sent\n- The design is extensible to support transport-specific notification methods\n\n## Transport-Specific Considerations\n\nWhile the core specification is transport-agnostic, implementations may support transport-appropriate notification mechanisms. Each route `type` must be separately specified:\n\n| Type | Transport | SEP |\n|----------|---------------|---------------------|\n| webhook | HTTP | (future SEP) |\n| signal | STDIO | (future SEP) |\n| message | WebSocket | (future SEP) |\n\n## Future Work\n\n- **Advanced triggers**\n - Conditional triggers based on result content\n - Time-based triggers\n - Aggregate triggers across multiple tool calls\n- **Delivery guarantees**\n - Retry policies\n - Delivery confirmation\n - Dead letter queues\n- **Security enhancements**\n - Notification authentication\n - Encrypted payloads\n - Rate limiting\n\n## Alternatives\n\n- **Server-side configuration only**\nRejected because it reduces flexibility and requires server administrators to manage client-specific configurations.\n- **Extension of progress notifications**\nRejected because progress notifications serve a different purpose and have different delivery requirements.\n- **Transport-specific specifications**\nRejected because it would fragment the ecosystem and make it harder to build transport-agnostic tools.\n\n## Backwards Compatibility\n\nThis feature is fully backward compatible:\n\n- Clients must opt in via the `notificationConfig` capability\n- The configuration parameter is optional\n- Servers that don't support the feature will ignore the configuration\n- Existing tool call behavior remains unchanged\n\n## Security Implications\n\n- Notification endpoints must be validated by servers\n- Authentication credentials in configurations must be handled securely\n- Rate limiting should be implemented to prevent abuse\n- Sensitive data may need to be filtered from notifications based on security policies","author":{"url":"https://github.com/Joffref","@type":"Person","name":"Joffref"},"datePublished":"2025-07-17T15:06:18.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":13},"url":"https://github.com/992/modelcontextprotocol/issues/992"}
| route-pattern | /_view_fragments/issues/show/:user_id/:repository/:id/issue_layout(.:format) |
| route-controller | voltron_issues_fragments |
| route-action | issue_layout |
| fetch-nonce | v2:ca0c82df-4ba1-0846-1027-0cd0a11c65a7 |
| current-catalog-service-hash | 81bb79d38c15960b92d99bca9288a9108c7a47b18f2423d0f6438c5b7bcd2114 |
| request-id | D3F6:2C32F8:EDCCA:132635:6A5BE68C |
| html-safe-nonce | 6102e90a7c75f242d36e2ca2f7cbba969421f1fb8e4f4e73b1808ba57e2dbd04 |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiJEM0Y2OjJDMzJGODpFRENDQToxMzI2MzU6NkE1QkU2OEMiLCJ2aXNpdG9yX2lkIjoiOTAwOTE1Njc1MjI5NTA2MTEzMiIsInJlZ2lvbl9lZGdlIjoiaWFkIiwicmVnaW9uX3JlbmRlciI6ImlhZCJ9 |
| visitor-hmac | 80085ddc1686025db727af36a4b72a97cd28f86f4ca89c57cbb9045827160048 |
| hovercard-subject-tag | issue:3239936538 |
| github-keyboard-shortcuts | repository,issues,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | / |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/_view_fragments/issues/show/modelcontextprotocol/modelcontextprotocol/992/issue_layout |
| twitter:image | https://opengraph.githubassets.com/f59dc7750292ee5a7349828dfee19ac156945f786fda3638f9b4d3e44cb2cd21/modelcontextprotocol/modelcontextprotocol/issues/992 |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/f59dc7750292ee5a7349828dfee19ac156945f786fda3638f9b4d3e44cb2cd21/modelcontextprotocol/modelcontextprotocol/issues/992 |
| og:image:alt | Preamble Authors: Mathis Joffre Abstract This proposal describes a mechanism for configuring notification routing when tool calls complete. Using this mechanism: Clients can specify where and how t... |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| og:author:username | Joffref |
| hostname | github.com |
| expected-hostname | github.com |
| None | 5290d7e14309ad1e76106a9c4237bd1041517e83ea182c8ab756752cb0c6940b |
| turbo-cache-control | no-preview |
| go-import | github.com/modelcontextprotocol/modelcontextprotocol git https://github.com/modelcontextprotocol/modelcontextprotocol.git |
| octolytics-dimension-user_id | 182288589 |
| octolytics-dimension-user_login | modelcontextprotocol |
| octolytics-dimension-repository_id | 862570523 |
| octolytics-dimension-repository_nwo | modelcontextprotocol/modelcontextprotocol |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 862570523 |
| octolytics-dimension-repository_network_root_nwo | modelcontextprotocol/modelcontextprotocol |
| turbo-body-classes | logged-out env-production page-responsive |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | 9c975978430e9ad293956f2bbdaf153b1bd84a99 |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
Links:
Viewport: width=device-width