Anthropic message threads (#11545)

Anthropic can keep a conversation on the server (message threads beta).
A request can then continue from the previous reply and send only the
new messages, instead of the whole history every turn.

pi uses this when a session id is set and caching is on. It still builds
the full request, compares it with the previous one, and only sends the
new part if nothing earlier changed.

- Default: on only for the anthropic provider on api.anthropic.com.
  Gateways that drop unknown fields would lose the conversation.
  Override with compat.supportsMessageThreads.
- If the server rejects a continuation, pi resends the full request
  as a new thread. Other errors turn threads off for the session.
- New option StreamOptions.disableContinuation. The cache warmer sets
  it so warm requests don't replace the session's thread.
- cleanupSessionResources(sessionId) frees a session's thread state.

packages/coding-agent/CHANGELOG.md

@@ -5,6 +5,7 @@
 ### Added
 
 - `pi update` now lists each updated package with its npm version or git commit change, for example `npm:foo 1.2.0 -> 1.3.0`, and prints `Packages are already up to date` when nothing changed. When nothing changed but an update printed a warning, for example because npm left the old version installed after resolving stale registry metadata, it prints `No packages were updated, see the warnings above` instead; in offline mode it prints `Skipped package updates (offline mode)`. `PackageManager.update()` reports each changed package as a new `result` progress event whose `message` holds that line ([#11707](https://github.com/earendil-works/pi/pull/11707), [#11711](https://github.com/earendil-works/pi/pull/11711))
+- Anthropic sessions now use message threads: each request sends only the new messages and the server continues from the previous reply, falling back to the full request when the conversation changed (for example after compaction or a tool list change). On by default for the `anthropic` provider on `api.anthropic.com`; set `compat.supportsMessageThreads` in `models.json` to enable it for gateways that forward the `thread` field, or to turn it off. Cache warming does not touch the session's thread ([#11545](https://github.com/earendil-works/pi/pull/11545))
 
 ## [1.5.0] - 2026-11-02
 

packages/coding-agent/docs/extensions.md

@@ -878,6 +878,8 @@ pi.on("before_provider_request", (event, ctx) => {
 
 This is mainly useful for debugging provider serialization and cache behavior.
 
+The payload is the full request. Providers that continue from a previous response (OpenAI Codex over WebSocket, Anthropic message threads) may send only the part that is new since the previous request.
+
 #### after_provider_response
 
 Fired after an HTTP response is received and before its stream body is consumed. Handlers run in extension load order.

packages/coding-agent/docs/models-configuration.md

@@ -172,6 +172,7 @@ Also applies to `azure-openai-responses` and `openai-codex-responses`.
 | `supportsMidConvoEffort` | boolean | `false` | Whether the exact model transport supports effort-only system messages and thinking binding controls. |
 | `supportsMidConvoSystemMessages` | boolean | `false` | Whether the exact model accepts system-role messages inside the conversation. When false, later system messages are folded into the top-level system prompt. |
 | `supportsMidConvoToolChanges` | boolean | `false` | Whether the exact model accepts mid-conversation tool_addition and tool_removal blocks. Requires supportsMidConvoSystemMessages. |
+| `supportsMessageThreads` | boolean |  | Whether the endpoint supports Anthropic message threads, which let a request with a session id send only the new messages and continue from the previous reply. Default: true for provider anthropic with a baseUrl on api.anthropic.com, false otherwise. Set true for gateways that forward the thread request field. |
 | `allowedFallbackModels` | `object[]` |  | Models Anthropic accepts for server-side refusal fallback, with local pricing metadata for returned fallback responses. When absent or empty, callers must omit fallbacks; Anthropic rejects the field for models with no permitted fallback targets. |
 
 ### `bedrock-converse-stream`

packages/coding-agent/schemas/models.schema.json

@@ -1081,6 +1081,10 @@
           "description": "Whether the exact model accepts mid-conversation tool_addition and tool_removal blocks. Requires supportsMidConvoSystemMessages.",
           "default": false
         },
+        "supportsMessageThreads": {
+          "type": "boolean",
+          "description": "Whether the endpoint supports Anthropic message threads, which let a request with a session id send only the new messages and continue from the previous reply. Default: true for provider anthropic with a baseUrl on api.anthropic.com, false otherwise. Set true for gateways that forward the thread request field."
+        },
         "allowedFallbackModels": {
           "type": "array",
           "items": {

packages/coding-agent/src/core/cache-warmer.ts

@@ -333,6 +333,8 @@ export class CacheWarmer {
 				.streamSimple(run.model, run.context, {
 					...run.options,
 					maxTokens: 1,
+					// The warm request repeats the last request; it must not replace the conversation's thread state.
+					disableContinuation: true,
 					maxRetries: 0,
 					signal: run.controller.signal,
 				})

packages/coding-agent/test/cache-warmer.test.ts

@@ -155,7 +155,14 @@ describe("cache warming", () => {
 
 		expect(calls[0]).toMatchObject({
 			model: adaptiveModel,
-			options: { reasoning: "high", sessionId: "s", transformHeaders, maxTokens: 1, maxRetries: 0 },
+			options: {
+				reasoning: "high",
+				sessionId: "s",
+				transformHeaders,
+				maxTokens: 1,
+				maxRetries: 0,
+				disableContinuation: true,
+			},
 		});
 		expect(calls[0].options?.signal).not.toBe(signal);
 		expect(events[0]).toMatchObject({