Repository navigation
ci: Version Packages - #2213
Open
github-actions[bot] wants to merge 1 commit into
Open
ci: Version Packages#2213github-actions[bot] wants to merge 1 commit into
github-actions[bot] wants to merge 1 commit into
Conversation
Contributor
|
Important Review skippedBot user detected. To trigger a single review, invoke the
|
github-actions
Bot
force-pushed
the
changeset-release/main
branch
4 times, most recently
from
November 21, 2026 05:08
6145e33 to
aece3d6
Compare
github-actions
Bot
force-pushed
the
changeset-release/main
branch
from
November 21, 2026 06:43
aece3d6 to
b0bbe24
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.
Releases
@tanstack/ai@1.9.0
Minor Changes
f25954a- Add optimization-driven, deterministic tool-result context projection.@tanstack/aigains acontextfield (ToolContextPolicy) on tooldefinitions plus the
FieldProjectiontype and its combinators(
truncateHead,truncateTail,truncateHeadTail,dropWithRef,aboveTokens).FieldProjectionis branded, so a projection is always builtwith a combinator, never a bare object. The core reads nothing from it (same
posture as
replay); behavior lives in the new package.@tanstack/ai-optimize: achat()middleware that truncates onlyoptimized tool-result (and
optimize()-declared argument) fields in themodel projection, never in the canonical messages, event stream, or UI.
Deterministic strategies only, a run-scoped stash, an injected
retrieve_contexttool, and anoptimize()wrapper for foreign/MCP toolsets. The combinators are re-exported here too.
Patch Changes
e5bb4d5- The lazy tool discovery tool no longer saysUnknown toolfor a tool that the model can already call. Before, it knew only the lazy tools. When a model asked it for an eager tool, or for a tool that a middleware added, it returnedUnknown tool, and some models then stopped using that tool. Discovery now returns the definition of each tool offered to the model in that step. An eager tool that a middleware removed still getsUnknown tool. A lazy tool that a middleware removed still returns its definition.@tanstack/ai-cartesia@0.1.0
Minor Changes
5aa5808- Add Cartesia text-to-speech and batch transcription adapters.Patch Changes
e5bb4d5,f25954a]:@tanstack/ai-computer-sandbox@1.2.0
Minor Changes
#1982
d595931- Add thepersist.reconnectoption tosandboxComputer(). It keeps the sandbox of each thread and reconnects to the same sandbox later withprovider.resume, so a user who comes back to a thread gets the same desktop. A kept sandbox is not destroyed when its thread closes or goes idle.#1982
d595931-sandboxComputer()takes apersist.snapshotsoption. The driver saves the files of one folder, like a browser profile, when a thread closes or goes idle. It restores them into the next new sandbox of the thread, so a sign-in survives a sandbox that is gone. A kept sandbox that comes back with an empty disk gets them back too. Saves and restores of one thread run one at a time. This works with every provider. CallsaveComputerSnapshot({ sandbox, threadId, snapshots })to save right away, for example after a sign-in.@tanstack/ai-sandboxexportscaptureSandboxFilesandrestoreSandboxFiles, the file capture and restore of portable snapshots.Patch Changes
d595931]:@tanstack/ai-harness@0.5.0
Minor Changes
#2145
ca20519- Routed agents now remember their own tool calls, you can pass your own handoff tool, and aharnessAgentcan be a root agent.routing.historysets what each model call gets of the routed runs of earlier turns. With'own'(the default), a routed agent gets its own earlier runs in full, with tool calls and results. It gets the other agents and the main model as labeled text in a user message, for examplemain:\n.... The main model gets the routed answers the same way, so no model reads the words of another speaker as its own.'all'gives every speaker every routed run in full.'text'gives routed runs as text only, as before.routing.handoffToolmakes the handoff tool of each speaker. It getsfrom,targets,allowHuman, andhandoff. The tool can have any name and its own fields.harnessAgentinagentswithroutingkeeps one child session across turns. Each turn sends it what happened in the thread since its last turn. Its model gets thehandofftool, so it can hand back.The default
history: 'own'changes the messages that routed agents and handoff targets get. A target gets the turns of the main model as labeled text. To keep the earlier messages, and the prompt cache of the provider for a handoff target, setrouting: { router, history: 'text' }.An app tool with the name of the handoff tool now replaces the harness tool. Before, the turn failed with a duplicate tool error.
A handoff from the
handofftool of a root agent now adds the note messages to the thread, as a router or code handoff does. Before, the next speaker did not see that the agent handed off. On models that reject a context that ends on an assistant message (new Claude models), the main model failed after a hand-back.Patch Changes
#2175
3e9a35d- A log that gives its ownership back when its sessions have no work now also clears its pending snapshot timer, so the timer no longer stays armed for a snapshot this host cannot write.Updated dependencies [
e5bb4d5,f25954a]:@tanstack/ai-optimize@0.1.0
Minor Changes
f25954a- Add optimization-driven, deterministic tool-result context projection.@tanstack/aigains acontextfield (ToolContextPolicy) on tooldefinitions plus the
FieldProjectiontype and its combinators(
truncateHead,truncateTail,truncateHeadTail,dropWithRef,aboveTokens).FieldProjectionis branded, so a projection is always builtwith a combinator, never a bare object. The core reads nothing from it (same
posture as
replay); behavior lives in the new package.@tanstack/ai-optimize: achat()middleware that truncates onlyoptimized tool-result (and
optimize()-declared argument) fields in themodel projection, never in the canonical messages, event stream, or UI.
Deterministic strategies only, a run-scoped stash, an injected
retrieve_contexttool, and anoptimize()wrapper for foreign/MCP toolsets. The combinators are re-exported here too.
Patch Changes
e5bb4d5,f25954a]:@tanstack/ai-sandbox@1.6.0
Minor Changes
#1982
d595931-sandboxComputer()takes apersist.snapshotsoption. The driver saves the files of one folder, like a browser profile, when a thread closes or goes idle. It restores them into the next new sandbox of the thread, so a sign-in survives a sandbox that is gone. A kept sandbox that comes back with an empty disk gets them back too. Saves and restores of one thread run one at a time. This works with every provider. CallsaveComputerSnapshot({ sandbox, threadId, snapshots })to save right away, for example after a sign-in.@tanstack/ai-sandboxexportscaptureSandboxFilesandrestoreSandboxFiles, the file capture and restore of portable snapshots.Patch Changes
3e9a35d,e5bb4d5,f25954a,ca20519]:@tanstack/ai-acp@1.0.4
Patch Changes
d595931,3e9a35d,e5bb4d5,f25954a,ca20519]:@tanstack/ai-code-mode@1.1.3
Patch Changes
3e9a35d,e5bb4d5,f25954a,ca20519]:@tanstack/ai-computer-use@1.1.3
Patch Changes
3e9a35d,e5bb4d5,f25954a,ca20519]:@tanstack/ai-harness-cli@0.1.4
Patch Changes
3e9a35d,e5bb4d5,1a23e34,f25954a,ca20519]:@tanstack/ai-mcp@1.3.2
Patch Changes
#2150
1a23e34-createMCPEventWebhookno longer requires thewebhook-idheader to equal the event'seventId. SEP-3415 sayswebhook-idis a per-message id that must not be derived fromeventId, so the receiver returned 400 for every event from a server that follows the spec. It now accepts both a per-message id and the olderwebhook-id=eventIdform.onEventstill gets the header asmessageId, so keep deduplicating on(subscription.id, messageId).Updated dependencies [
3e9a35d,e5bb4d5,f25954a,ca20519]:@tanstack/ai-openai@1.2.1
Patch Changes
#2208
b8a0ce6-openaiImage()acceptsgpt-image-2.6-flareandgpt-image-2.6-sunburst. Both take sizes1024x1024,1536x1024,1024x1536andauto, andqualitytakeslow,medium,high,xhigh,maxandauto. Before, both ids threwUnknown image modelbefore any request was sent.Updated dependencies [
e5bb4d5,f25954a]:@tanstack/ai-sandbox-cloudflare@1.1.1
Patch Changes
#1982
d595931- Run eachexecandspawncommand in a subshell. Cloudflare runs every command in one shared shell session, so a command that calledexitended the session and every later command failed. This broke the@tanstack/ai-computer-sandboxdriver on Cloudflare.Updated dependencies [
d595931,e5bb4d5,dfbacd1,f25954a]:@tanstack/ai-sandbox-docker@1.1.1
Patch Changes
#1982
d595931-publishPortsnow publishes each port on 127.0.0.1 only, andports.connectreturns ahttp://127.0.0.1:<port>address. Before, Docker published the ports on every network interface, so other computers on the network could connect.Updated dependencies [
d595931]:@tanstack/ai-sandbox-local-process@1.0.1
Patch Changes
#2162
dfbacd1-exec()no longer corrupts multi-byte UTF-8 characters instdoutandstderr. A character that a pipe split across two reads used to decode as U+FFFD. Output is now decoded across chunks, likespawn()already does.Updated dependencies [
d595931]:@tanstack/ai-skills@1.0.4
Patch Changes
3e9a35d,e5bb4d5,f25954a,ca20519]:@tanstack/ai-optimize-evals@0.0.1
Patch Changes
e5bb4d5,b8a0ce6,f25954a]: