Add native Harmony chat and sequential tool calling by sayanshaw24 · Pull Request #973 · microsoft/foundry-local

Sayan Shaw (sayanshaw24) · 2026-08-10T23:51:31Z

Summary

Adds native GPT-OSS Harmony chat and sequential tool calling. Targets main; the shared validation prerequisite #1302 is now merged.

The implementation separates model protocol selection from ordinary tool framing and replaces the prototype parser while preserving the original contributors' commits.

What changes

Read assistant headers and message endings using token IDs rather than marker-text matching.
Keep source code, schemas, arguments, and tool results as literal prompt data.
Reuse shared schema validation, bounded recovery, and session handling.
Preserve raw and JSON tool-body encoding through streaming, hosting, and stored Responses replay.
Reject native Harmony loading when the required runtime or model-template support is unavailable.

Related changes and runtime requirements

Uses the shared batch validation and JSON grammar from Share tool-call batch validation and JSON grammar #1302, now merged into main.
The terminal-token API is merged in Report which EOS token ended an Engine turn onnxruntime-genai#2895. Native Harmony still requires a runtime package containing that API; the currently pinned older package does not provide it. Ordinary models retain build compatibility with older headers and libraries.
The corresponding model-template change is Add a GPT-OSS chat template patch for tool-call history olive-recipes#695.

Validation

The reviewed snapshot builds with the matching local runtime.
All 56 local contract tests and 917 existing regression tests passed; one optional Qwen-package fixture skipped.
The guarded runtime code also compiled with the actual older pinned headers.
Subsequent combined-stack qualification passed all 14 supported Harmony HTTP gates.

These are previously completed validations; updating the branch does not imply a new GPU qualification run.

Latest main and CI refresh

Merged current main after Share tool-call batch validation and JSON grammar #1302 landed, preserving the existing branch history.
Fixed the two Windows x64/ARM64 MSVC shadowing diagnostics by renaming a parameter and a local variable, without changing protocol or runtime behavior.
The old missing dependency-regressions XML was a consequence of the failed Windows build, not a separate test failure.
The refreshed branch built with the terminal-token-capable runtime; all 627 selected Harmony, transcript/replay, tool, grammar, template, session, and Responses regression tests passed on CPU, with no skips.
Windows x64/ARM64 confirmation remains dependent on the new CI run; Linux x64/ARM64 and macOS ARM64 had passed in the inspected old run.

Scope

Sequential tool use. The upstream Harmony template renders one tool call per assistant message, so true parallel calls remain outside this PR.

vercel · 2026-08-10T23:51:37Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
foundry-local	Ready	Preview	Nov 12, 2026 12:13am UTC

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Build on the original Harmony PR protocol-selection idea while keeping regular tool calling unchanged. Qualify native token/template contracts, decode channels and handoffs, protect transcript data, and reuse shared admission and recovery. Generation is sequential; coding-model qualification remains blocked. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

The native implementation supersedes the original four-file accumulator prototype. Keep its commits and authorship without retaining a second Harmony parser. This merge deliberately leaves the new implementation tree unchanged. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Carry per-call encoding outside public JSON through response publication, storage, replay snapshots and compaction. Keep raw-envelope callers compatible using the same metadata map. Recovered streaming items use the completed turn context so their encoding matches final responses and history. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Qualify the GPT-OSS thinking-only tool-call shape and allow runtimes with terminal-token identity to use Harmony through the Generator backend. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

🟡 Changes recommended

Qualified Harmony models can enter the unsupported classic Generator path without required Engine terminal-token metadata, while decoder and native call behavior lack checked-in regression tests.

2 open issues

What changed in this PR

Adds native GPT-OSS Harmony support to Foundry Local’s chat pipeline, preserving tool-body encoding across hosting and Responses replay. The PR also expands predictive SDK sessions, embedding usage reporting, and regression coverage.

Changes:

Adds token-ID-based Harmony decoding, protected prompt construction, native guidance, and runtime capability checks.
Exposes ranking and typed-decision sessions in C++ and Python, with runtime lifetime and cancellation handling.
Updates embedding token accounting, hosting reliability, documentation, tests, and CI model staging.

File	Description
`sdk_v2/rust/tests/http_extract_test.rs`	Consolidates extraction checks and relocates test artifacts.
`sdk_v2/python/test/unit/test_non_generative_sessions.py`	Tests predictive contracts, validation, and handle cleanup.
`sdk_v2/python/test/unit/test_model_download_cancellation.py`	Parameterizes download cancellation coverage.
`sdk_v2/python/test/unit/test_imports.py`	Checks new predictive API exports.
`sdk_v2/python/test/integration/test_non_generative_sessions.py`	Adds native ranking and decision integration tests.
`sdk_v2/python/test/conftest.py`	Adds cached predictive-model fixtures.
`sdk_v2/python/src/foundry_local_sdk/session.py`	Implements typed ranking and decision sessions.
`sdk_v2/python/src/foundry_local_sdk/session_types.py`	Defines predictive request and response types.
`sdk_v2/python/src/foundry_local_sdk/__init__.py`	Exports predictive sessions and types.
`sdk_v2/python/README.md`	Documents predictive APIs and lifecycle.
`sdk_v2/python/examples/ranking_and_decision.py`	Demonstrates predictive session usage.
`sdk_v2/js/test/openai-client-dispose.test.ts`	Tests request-scoped session release.
`sdk_v2/INFERENCING.md`	Documents image input formats and limitations.
`sdk_v2/cs/test/FoundryLocal.Tests/ManagerLifetimeTests.cs`	Consolidates lease tests with bounded waits.
`sdk_v2/cpp/test/sdk_api/shared_test_env.h`	Allows explicit embedding-model selection.
`sdk_v2/cpp/test/sdk_api/responses_vision_test.cc`	Exercises standard image data URLs.
`sdk_v2/cpp/test/sdk_api/non_generative_test.cc`	Adds predictive SDK integration coverage.
`sdk_v2/cpp/test/sdk_api/embeddings_test.cc`	Tests usage accounting and cancellation isolation.
`sdk_v2/cpp/test/internal_api/web_service_test.cc`	Checks predictive session closure preserves loaded models.
`sdk_v2/cpp/test/internal_api/non_generative_runtime_test.cc`	Tests predictive queuing, shutdown, and dispatch.
`sdk_v2/cpp/test/internal_api/non_generative_contract_test.cc`	Tests embedded-NUL rejection.
`sdk_v2/cpp/test/internal_api/ipc_channel_test.cc`	Covers socket startup and unlink races.
`sdk_v2/cpp/test/internal_api/host_lifetime_test.cc`	Consolidates reconnect-deadline coverage.
`sdk_v2/cpp/test/internal_api/bootstrap_descriptor_test.cc`	Synchronizes owner-removal assertions.
`sdk_v2/cpp/test/internal_api/azure_catalog_test.cc`	Tests catalog metadata fallback and precedence.
`sdk_v2/cpp/test/CMakeLists.txt`	Registers predictive tests and IPC instrumentation.
`sdk_v2/cpp/src/service/responses_handler.cc`	Stores generated tool-call encodings.
`sdk_v2/cpp/src/service/non_generative_handlers.cc`	Returns HTTP 503 for runtime capacity exhaustion.
`sdk_v2/cpp/src/service/embeddings_handler.cc`	Returns measured embedding token usage.
`sdk_v2/cpp/src/model.h`	Declares loaded predictive-runtime acquisition.
`sdk_v2/cpp/src/model.cc`	Adds predictive runtime acquisition and I/O descriptors.
`sdk_v2/cpp/src/manager.cc`	Drains predictive requests during destruction.
`sdk_v2/cpp/src/inferencing/session/types.h`	Adds native raw and JSON call encodings.
`sdk_v2/cpp/src/inferencing/session/session.h`	Adds request-lifetime and model-binding hooks.
`sdk_v2/cpp/src/inferencing/session/session.cc`	Dispatches predictive sessions and acquires request leases.
`sdk_v2/cpp/src/inferencing/predictive/session_executor.h`	Extracts bounded, cancellation-aware runtime execution.
`sdk_v2/cpp/src/inferencing/predictive/non_generative_session.h`	Declares predictive session lifecycle state.
`sdk_v2/cpp/src/inferencing/predictive/non_generative_session.cc`	Implements predictive JSON dispatch and runtime leases.
`sdk_v2/cpp/src/inferencing/predictive/non_generative_runtime.h`	Adds runtime interfaces and capacity errors.
`sdk_v2/cpp/src/inferencing/predictive/non_generative_runtime.cc`	Uses the cancellation-aware executor.
`sdk_v2/cpp/src/inferencing/model_load_manager.h`	Declares predictive lifecycle coordination.
`sdk_v2/cpp/src/inferencing/model_load_manager.cc`	Implements predictive leases and shutdown draining.
`sdk_v2/cpp/src/inferencing/generative/toolcalling/tool_call_context.h`	Tracks Harmony protocol and body encoding.
`sdk_v2/cpp/src/inferencing/generative/toolcalling/grammar.cc`	Builds native Harmony guidance.
`sdk_v2/cpp/src/inferencing/generative/openresponses/response_store.h`	Stores per-call encoding metadata.
`sdk_v2/cpp/src/inferencing/generative/openresponses/response_store.cc`	Preserves encoding metadata through storage and replay.
`sdk_v2/cpp/src/inferencing/generative/openresponses/response_converter.cc`	Restores call encodings and forwards reasoning effort.
`sdk_v2/cpp/src/inferencing/generative/openresponses/response_chain.h`	Carries encoding metadata in replay chains.
`sdk_v2/cpp/src/inferencing/generative/genai_model_instance.h`	Exposes qualified Harmony capabilities.
`sdk_v2/cpp/src/inferencing/generative/genai_model_instance.cc`	Qualifies Harmony tokenizer, template, and runtime support.
`sdk_v2/cpp/src/inferencing/generative/embeddings/embeddings_session.h`	Threads usage accounting through embedding generation.
`sdk_v2/cpp/src/inferencing/generative/embeddings/embeddings_session.cc`	Counts prepared embedding input tokens.
`sdk_v2/cpp/src/inferencing/generative/chat/search_options.cc`	Enables Harmony guidance and host-side stops.
`sdk_v2/cpp/src/inferencing/generative/chat/prepared_chat_prompt.h`	Records Harmony continuation metadata.
`sdk_v2/cpp/src/inferencing/generative/chat/prepared_chat_prompt.cc`	Builds protected prompts and validates continuations.
`sdk_v2/cpp/src/inferencing/generative/chat/onnx_engine_chat_stream.cc`	Propagates terminal-token identity into turn usage.
`sdk_v2/cpp/src/inferencing/generative/chat/onnx_chat_engine.h`	Adds terminal-token metadata to results.
`sdk_v2/cpp/src/inferencing/generative/chat/onnx_chat_engine.cc`	Reads terminal-token identity from Engine events.
`sdk_v2/cpp/src/inferencing/generative/chat/harmony_protocol.h`	Defines Harmony controls and continuation state.
`sdk_v2/cpp/src/inferencing/generative/chat/harmony_decoder.h`	Declares decoder events and recovery interface.
`sdk_v2/cpp/src/inferencing/generative/chat/harmony_decoder.cc`	Parses native headers, bodies, and completion tokens.
`sdk_v2/cpp/src/inferencing/generative/chat/chat_transcript.h`	Clarifies native raw-body replay comments.
`sdk_v2/cpp/src/inferencing/generative/chat/chat_transcript.cc`	Clarifies custom argument normalization comments.
`sdk_v2/cpp/src/inferencing/generative/chat/chat_template.h`	Declares Harmony prompt projection.
`sdk_v2/cpp/src/inferencing/generative/chat/chat_template.cc`	Projects protected Harmony history and tool schemas.
`sdk_v2/cpp/src/inferencing/generative/chat/chat_generator.h`	Adds terminal-token identity to turn usage.
`sdk_v2/cpp/src/hosting/model_dispatcher.cc`	Adjusts hosted session creation and lock scope.
`sdk_v2/cpp/src/hosting/item_codec.cc`	Accepts native tool-call encodings over IPC.
`sdk_v2/cpp/src/hosting/ipc_channel_posix.cc`	Handles socket startup races before authentication.
`sdk_v2/cpp/src/contracts/non_generative.cc`	Rejects ranking answers containing embedded NULs.
`sdk_v2/cpp/src/contracts/embeddings.h`	Widens token counts to 64 bits.
`sdk_v2/cpp/src/contracts/chat_completions_converter.cc`	Forwards explicit reasoning effort.
`sdk_v2/cpp/src/catalog/azure_catalog_models.cc`	Adds capability-tag fallback and generic integer parsing.
`sdk_v2/cpp/src/api/hosted_manager.cc`	Reserves model selection during hosted session creation.
`sdk_v2/cpp/README.md`	Documents predictive sessions and native Harmony requirements.
`sdk_v2/cpp/include/foundry_local/foundry_local_cpp.inline.h`	Implements predictive JSON convenience methods.
`sdk_v2/cpp/include/foundry_local/foundry_local_cpp.h`	Exposes predictive sessions and handles `GetMessage` conflicts.
`sdk_v2/cpp/CMakeLists.txt`	Probes terminal-token support and includes new sources.
`.pipelines/templates/steps-build-linux.yml`	Stages and runs required predictive integration models.
`.pipelines/templates/stages-python.yml`	Stages and verifies predictive test packages.
`.pipelines/templates/fetch-models-from-artifacts-feed.yml`	Adds CLM and KEV artifact downloads.
`.github/instructions/cpp-tool-calling.instructions.md`	Documents Harmony token and completion invariants.
`.github/instructions/cpp-testing.instructions.md`	Documents embedding-model test overrides.

_{Review effort: Balanced}

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Sayan Shaw (sayanshaw24) · 2026-11-09T21:27:49Z

67a8db7 addresses the additional review-summary finding: declarations and historical calls with whitespace or metadata separators in their names are rejected with INVALID_ARGUMENT before Harmony template rendering. Coverage exercises all six ASCII whitespace separators plus both internal metadata separator bytes. The leading-instruction and custom-input normalization fixes remain in place. Expanded combined native/Harmony selection passed 291 tests with one optional package skip, and the original supported Harmony HTTP matrix passed 14/14 again. Native recipient selection and protocol behavior for valid names are unchanged.

Copilot

🔵 Needs a closer look

Human assessment is needed because transcript projection and stored-input replay can lose historical tool-call encoding distinctions, while the Harmony grammar branch lacks regression coverage.

No open issues 🎉

_{Review effort: Balanced}

Carry known per-call encoding metadata through stored input replay as well as output replay. Cover native JSON versus raw history bodies, choice-specific headers, reserved boundary IDs, and native patch grammar admission. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Sayan Shaw (sayanshaw24) · 2026-11-09T22:02:30Z

Addressed the latest provenance/coverage concerns in 29d8b3e. Stored-input reconstruction now receives the same trusted per-call encoding map already used for generated output reconstruction. Tests round-trip structured, native-JSON, and native-raw custom bodies through stored input and stored output, verifying the historical encoding, tool kind, exact payload, and rendered raw-vs-JSON body even when the current declaration admits raw bodies. Added Harmony grammar tests for automatic/required/named/none choice with and without reasoning, both legal recipient header orders, authoritative boundary IDs and EOS exclusion, native stock-patch bodies, and explicit rejection of unsupported native custom formats. Final combined-stack suites passed 518 tests against both updated GenAI and the exact pinned 0.19.0 runtime (one optional package test skipped each). Public caller payloads do not acquire trusted provenance from client-supplied fields.

Sayan Shaw (sayanshaw24) · 2026-11-09T22:02:56Z

Final post-fix combined-stack regression also passed all 14 supported native Harmony HTTP gates, including streaming/non-streaming required/named choice, continuation, serial calls, same-tool/different-arguments, and failed-result no-retry. The shared stack additionally passed 16/16 HTTP gates on each of three native-protocol configurations. No prompts/assertions were relaxed. Combined integration head 63d28ef is byte-identical to the tested SDK tree, and all test-owned model servers have been stopped.

Copilot

🔵 Needs a closer look

Human assessment is needed because later grammar changes can alter historical custom-call encodings, including during stored-response replay.

No open issues 🎉

_{Review effort: Balanced}

Historical generated and stored canonical/JSON/raw calls keep their original body encoding when declarations change. Only unannotated caller history infers the current declaration, preserving existing caller behavior. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Sayan Shaw (sayanshaw24) · 2026-11-09T22:22:29Z

Follow-up 63f98e0 addresses the remaining historical-format distinction explicitly. The transcript now distinguishes recorded generated/stored encoding from unannotated caller history. Recorded canonical, native-JSON, and native-raw bodies retain their original encoding even if the current custom grammar is added or removed. Only unannotated caller-supplied calls continue inferring the current declaration, preserving existing caller behavior. The stored-input/output regression now covers all three historical encodings with and without a current raw-body grammar, and a separate regression protects unannotated caller behavior. Focused combined-stack tests pass 519 cases (one optional package skip).

Copilot

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

Incorporate the merged shared tool-validation prerequisite and current main without rewriting branch history. Rename the raw-envelope parameter and assistant-prefix local rejected by MSVC warnings-as-errors; preserve all protocol behavior. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Sayan Shaw added 2 commits August 10, 2026 15:53

Add ToolCallConfig and header-inspection mode for Harmony tool calling

6836b3c

Add ToolCallConfig, header-inspection mode, and Harmony unit tests

2cdc513

vercel Bot deployed to Preview August 11, 2026 00:02

View deployment

Fix GCC -Werror=missing-field-initializers in ToolCallConfig::Simple()

7566cd3

vercel Bot deployed to Preview August 11, 2026 15:38

View deployment

Justin Chu (justinchuby) reviewed Aug 19, 2026

View reviewed changes

Comment thread sdk_v2/cpp/src/inferencing/generative/toolcalling/tool_call_config.h Resolved

Justin Chu (justinchuby) reviewed Aug 19, 2026

View reviewed changes

Comment thread sdk_v2/cpp/src/inferencing/generative/toolcalling/tool_call_config.h Resolved

Merge main into harmony, resolve chat_session.cc conflicts

31e7937

vercel Bot deployed to Preview August 19, 2026 18:40

View deployment

Baiju Meswani (baijumeswani) and others added 6 commits November 5, 2026 23:20

Share byte-preserving tool batch admission and structural JSON grammar

0d0da52

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Merge main into shared tool admission branch

aa76ad3

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Keep caller-supplied replay calls on the default encoding

21c6f0e

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1a563a06-04d5-47cf-9b20-af502d60b1eb

Baiju Meswani (baijumeswani) changed the base branch from main to baijumeswani/tool-batch-admission November 7, 2026 00:59

Baiju Meswani (baijumeswani) added this pull request to stack #1303 November 7, 2026 00:59

vercel Bot deployed to Preview November 7, 2026 00:59

View deployment

Support legal Harmony tool history

e81aca6

Qualify the GPT-OSS thinking-only tool-call shape and allow runtimes with terminal-token identity to use Harmony through the Generator backend. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

vercel Bot deployed to Preview November 7, 2026 19:37

View deployment

Merge main into Harmony support

4b07671

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

vercel Bot deployed to Preview November 8, 2026 03:01

View deployment

Sayan Shaw (sayanshaw24) marked this pull request as ready for review November 8, 2026 03:05

Copilot AI balanced review requested due to automatic review settings November 8, 2026 03:06

Sayan Shaw (sayanshaw24) mentioned this pull request Nov 8, 2026

Share tool-call batch validation and JSON grammar #1302

Merged

Copilot started reviewing on behalf of Sayan Shaw (sayanshaw24) November 8, 2026 03:06

View session

Copilot AI reviewed Nov 8, 2026

View reviewed changes

Comment thread sdk_v2/cpp/src/inferencing/generative/genai_model_instance.cc Resolved

Comment thread sdk_v2/cpp/src/inferencing/generative/chat/harmony_decoder.cc Resolved

Cover Harmony terminal and replay contracts

f9d2b43

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

vercel Bot deployed to Preview November 8, 2026 04:43

View deployment

vercel Bot deployed to Preview November 9, 2026 21:24

View deployment

Sayan Shaw (sayanshaw24) requested a balanced review from Copilot November 9, 2026 21:27

Copilot started reviewing on behalf of Sayan Shaw (sayanshaw24) November 9, 2026 21:28

View session

Copilot AI reviewed Nov 9, 2026

View reviewed changes

vercel Bot deployed to Preview November 9, 2026 22:02

View deployment

Sayan Shaw (sayanshaw24) requested a balanced review from Copilot November 9, 2026 22:02

Copilot started reviewing on behalf of Sayan Shaw (sayanshaw24) November 9, 2026 22:03

View session

Copilot AI reviewed Nov 9, 2026

View reviewed changes

Sayan Shaw (sayanshaw24) requested a balanced review from Copilot November 9, 2026 22:22

vercel Bot deployed to Preview November 9, 2026 22:22

View deployment

Copilot started reviewing on behalf of Sayan Shaw (sayanshaw24) November 9, 2026 22:23

View session

Copilot AI reviewed Nov 9, 2026

Copilot stopped reviewing on behalf of Sayan Shaw (sayanshaw24) due to an error November 9, 2026 22:23
Error performing PR review: Error: Failed to fetch hash algorithm for microsoft/foundry-local (status 401): HttpError: Bad credentials - https://docs.github.com/rest

Sayan Shaw (sayanshaw24) removed this pull request from stack #1303 November 10, 2026 01:46

Sayan Shaw (sayanshaw24) changed the title ~~Add Harmony (GPT-OSS) Tool Calling Support in FL~~ Add native Harmony chat and sequential tool calling Nov 10, 2026

Sayan Shaw (sayanshaw24) changed the base branch from baijumeswani/tool-batch-admission to main November 10, 2026 01:46

bmehta001 mentioned this pull request Nov 10, 2026

Support native tool continuations #1296

Open

Sayan Shaw (sayanshaw24) changed the base branch from main to baijumeswani/tool-batch-admission November 10, 2026 19:13

Sayan Shaw (sayanshaw24) added this pull request to stack #1319 November 10, 2026 19:13

Base automatically changed from baijumeswani/tool-batch-admission to main November 10, 2026 19:47

vercel Bot deployed to Preview November 10, 2026 19:55

View deployment

Merge main into native Harmony support

a90f5e7

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

vercel Bot deployed to Preview November 12, 2026 00:13

View deployment

Navigation Menu

Repository navigation

Add native Harmony chat and sequential tool calling - #973

Add native Harmony chat and sequential tool calling#973
Sayan Shaw (sayanshaw24) wants to merge 23 commits into
mainfrom
sayanshaw/harmony

Sayan Shaw (sayanshaw24) commented Aug 10, 2026 •

edited

Loading

Uh oh!

vercel Bot commented Aug 10, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

This branch was successfully deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

Sayan Shaw (sayanshaw24) commented Aug 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What changes

Related changes and runtime requirements

Validation

Latest main and CI refresh

Scope

Uh oh!

vercel Bot commented Aug 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🔵 Needs a closer look

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🔵 Needs a closer look

Uh oh!

Sayan Shaw (sayanshaw24) commented Nov 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

This branch was successfully deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Sayan Shaw (sayanshaw24) commented Aug 10, 2026 •

edited

Loading

vercel Bot commented Aug 10, 2026 •

edited

Loading