ref!: Remove plugins as part of public API by lforst · Pull Request #2386 · braintrustdata/braintrust-sdk-javascript

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) · 2026-08-19T17:57:15Z

ref braintrustdata/braintrust-sdk-javascript#1841

Removes Plugins and unsubscribe functionality.

Reasoning: Plugins are a bit of an unnecessary API to be exposed because they just attach stuff and un-attach stuff. The user can very easily just manage this by themselves. Also, unsubscribe is literally called from nowhere in the SDK.

…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

LangChain's `UsageMetadata` carries `output_token_details.reasoning`, and nothing in `js/src` or `integrations` read it. `completion_tokens` covers the spend, so what was lost is the breakdown: no way to see how much of a reasoning model's output went to thinking, on any LangChain or LangGraph span. Mapped `reasoning` only. `js/src/wrappers/mastra.ts:164-172` settles the scope, picking up the non-modality detail fields (`cacheRead`, `cacheWrite`, `reasoning`) and leaving the modality ones alone. `OutputTokenDetails` also carries `audio`, `image`, `video`, `document` and `text` through `ModalitiesTokenDetails`; the input side drops its copies too, so those are a separate change. Zero is recorded, not dropped, matching `prompt_cached_tokens: 0` in the existing snapshots. `js/src/instrumentation/plugins/langsmith-plugin.ts:456` has the same gap. Left alone because #2386 is moving that tree. Snapshot churn is 30 added lines across 10 files, every one `completion_reasoning_tokens`, both sides of each span-tree pair regenerated, no deletions. Ran `pnpm test` in `js/` (1648 passed), `check:typings`, `lint`, and `test:e2e:update` then `test:e2e` once each, not twice. On that assert pass `scenarios/git-metadata` timed out at 30s under parallel load. It passes in isolation here in 1.19s and on `main` in 1.20s, and the two scenarios this change touches pass, so I read it as local contention. Worth confirming on CI. Both new unit tests fail against `main` on assertions. --------- Co-authored-by: Dylan Pulver <dylanpulver@users.noreply.github.com> Co-authored-by: lforst <8118419+lforst@users.noreply.github.com>

Luca Forstner (lforst) added 4 commits August 19, 2026 17:57

ref!: Remove plugins as part of public API
5409d11

Merge origin/v4
7bf6fef

Update PR#2386
ee54d2b

Update PR#2386
4771547

Luca Forstner (lforst) mentioned this pull request Aug 20, 2026
v4 #1841
Open
10 tasks

Merge remote-tracking branch 'origin/v4' into lforst/dum-e/edinburgh-…
fb5c7cb
…ed36b547c4 # Conflicts: # js/src/exports.ts # js/src/instrumentation/index.ts # js/src/instrumentation/providers/eve-instrumentation.test.ts

Dylan Pulver (dylanpulver) mentioned this pull request Aug 21, 2026
fix(langchain): Record reasoning tokens as metric #2393
Merged

Luca Forstner (lforst) requested a review from Abhijeet Prasad (AbhiPrasad)August 24, 2026 12:02

Abhijeet Prasad (AbhiPrasad) approved these changes Aug 26, 2026
View reviewed changes

Luca Forstner (lforst) merged commit 55dd6c9 into v4Aug 26, 2026
92 of 94 checks passed

Luca Forstner (lforst) deleted the lforst/dum-e/edinburgh-ed36b547c4 branch August 26, 2026 18:09