feat(frontend): add DYN_ENABLE_NVEXT and DYN_ENABLE_FRONTEND_ADMIN_API master switches by nnshah1 · Pull Request #10556 · ai-dynamo/dynamo

nnshah1 · 2026-06-10T20:28:21Z

Summary

Adds two binary master switches that let operators close off non-OpenAI-spec surfaces on the frontend HTTP service without affecting inference, metrics, models, or health/liveness probes. Both default to true (no behavior change unless the operator explicitly opts out).

Env var	Default	Effect when false
`DYN_ENABLE_NVEXT`	`true`	chat/completions/responses handlers drop `request.nvext` at the boundary; routing-override headers (`x-worker-instance-id`, `x-prefill-instance-id`, `x-dp-rank`, `x-data-parallel-rank`, `x-prefill-dp-rank`) are ignored; response-side `extra_fields` opt-in returns the default empty selection
`DYN_ENABLE_FRONTEND_ADMIN_API`	`true`	`GET /busy_threshold` and `POST /busy_threshold` routes are not registered (404 instead of 503)

Each switch is also exposed as a HttpServiceConfig builder field (enable_nvext, enable_admin_api) so Python frontend code can flip it from CLI without an env var.

Motivation

PSIRT findings 6067559 / 6067561 / 6067562 / 6067563 / 6067566 flagged the frontend admin / metadata surfaces as unauthenticated. 6067564 documented the EPP↔frontend GAIE routing-header protocol (x-worker-instance-id etc.) as a by-design but operator-disable-able surface. The proposed deployment-guide remediation (DYN-2647) needed an in-process kill-switch so a default-secure deployment can lock these off without source patches.

DYN_ENABLE_NVEXT shuts off worker-targeting via header and body in one switch (header parsing + nvext field interpretation both gated by the same flag). DYN_ENABLE_FRONTEND_ADMIN_API shuts off the busy_threshold admin surface.

Compatibility

Defaults preserve current behavior end-to-end. Operators who set DYN_ENABLE_NVEXT=false should be aware:

EPP / GAIE serving breaks — the EPP needs nvext for routing
Prime-RL-style training breaks — relies on nvext.cache_salt for KV cache isolation across checkpoint steps
Multi-tenant agent platforms break — anything forwarding nvext.agent_hints / nvext.agent_context
Clients lose extra_fields response opt-in — nvext.extra_fields is silently ignored

Documented in the const docstring on DYN_ENABLE_NVEXT in environment_names.rs.

Design notes

Zero per-request perf impact — both flags read once at HttpService::run() (single env_is_falsey call), ANDed with the builder field, and stored as a plain bool on State. Handlers consume via state.nvext_enabled() — single struct-field load + predicted-taken branch. No atomic, no OnceLock, no per-request env lookup. Matches the existing precedent at service_v2.rs:698 (enable_rl_router).
Env-falsey only — unset env keeps enable_nvext=true from the builder. So a typo like DYN_ENABLE_NVEXT=ture falls through to true rather than silently disabling a default-on feature.
Admin-api gate is at the route-registration site — when disabled, the route literally doesn't exist (404), not a handler that 503s.
State::new API preserved — added State::new_with_flags(...) for the wiring; the original State::new(...) defaults nvext_enabled=true.

Test plan

cargo check -p dynamo-llm (clean)
cargo clippy -p dynamo-llm --no-deps (clean)
cargo fmt --check
Manual smoke: start frontend with DYN_ENABLE_NVEXT=false, send a chat request with x-worker-instance-id header → verify it's ignored
Manual smoke: start frontend with DYN_ENABLE_FRONTEND_ADMIN_API=false → verify GET /busy_threshold 404s

Out of scope (follow-ups)

CLI flag mirror in components/src/dynamo/frontend/frontend_args.py (--enable-nvext / --enable-frontend-admin-api)
Rename DYN_KV_INDEXER_TEST_ENDPOINTS → DYN_ENABLE_KV_INDEXER_TEST_API for naming parity (separate PR — touches a different component)
Secure-deployment-guide doc updates (covered under DYN-2647)

Summary by CodeRabbit

New Features
- Added configuration options to enable/disable nvext protocol extension processing (enabled by default)
- Added configuration option to control admin API endpoint availability (enabled by default)
Chores
- Added environment variables for dynamic runtime control of new configuration options

…I master switches Addresses PSIRT findings 6067559/6067562 (unauthenticated busy_threshold) and the 6067564 by-design routing-header surface by giving operators two binary master switches that fully close off the nvext extension protocol and the frontend admin API. DYN_ENABLE_NVEXT (default true) When false at frontend startup, the chat/completions/responses handlers drop request.nvext at the boundary and ignore the routing-override headers (x-worker-instance-id, x-prefill-instance-id, x-dp-rank, x-prefill-dp-rank). Response-side extra_fields opt-in is silently ignored. This shuts off worker-targeting via header AND body in one switch. Compatibility: disables EPP/GAIE serving, Prime-RL-style training that uses nvext.cache_salt for KV isolation, multi-tenant agent platforms that forward agent_hints/agent_context, and clients that opt into response disclosure via nvext.extra_fields. DYN_ENABLE_FRONTEND_ADMIN_API (default true) When false, the GET /busy_threshold and POST /busy_threshold routes are not registered. Inference, metrics, models, health, and liveness routes are unaffected. (The historical POST /clear_kv_blocks route was already unwired in #1629; not re-exposed.) Design notes - Both flags are read once at HttpService::run() via env_is_falsey on DYN_ENABLE_NVEXT / DYN_ENABLE_FRONTEND_ADMIN_API, ANDed with the HttpServiceConfig builder field. Env-falsey only — unset env keeps the default-true behavior intact. - State carries the nvext_enabled flag as a plain bool field set at construction; handlers read it via state.nvext_enabled() — single struct-field load on the hot path, no atomic, no per-request env lookup, no OnceLock. Matches the existing pattern used for StateFlags + the enable_rl_router gate around line 698. - The admin-api gate is purely at the route-registration site, so disabled mode means the route literally doesn't exist (404) rather than running a handler that 503s. Tested with cargo check and cargo clippy --no-deps on dynamo-llm. Signed-off-by: nnshah1 <neelays@nvidia.com>

devin-ai-integration

Devin Review found 1 potential issue.

View 3 additional findings in Devin Review.

coderabbitai · 2026-06-10T20:34:27Z

Walkthrough

This PR adds two master switches to the HTTP service layer. The nvext flag controls whether client-supplied routing-override extensions are processed by request handlers; the admin_api flag controls whether the busy_threshold admin routes are registered. Both are configurable via builder options and environment variables, with defaults set to enabled.

Changes

Feature flags for nvext extension and admin API routes

Layer / File(s)	Summary
Environment variable definitions `lib/runtime/src/config/environment_names.rs`	Introduces `DYN_ENABLE_NVEXT` and `DYN_ENABLE_FRONTEND_ADMIN_API` constants with documentation describing how falsy values disable routing-override header processing and admin API route registration.
State extension and initialization `lib/llm/src/http/service/service_v2.rs`	Adds `nvext_enabled: bool` field to `State`, refactors constructors to accept and store the flag via `State::new_with_flags()`, and provides `nvext_enabled()` accessor for request handlers.
Builder configuration and flag resolution `lib/llm/src/http/service/service_v2.rs`	Extends `HttpServiceConfig` with `enable_nvext` and `enable_admin_api` builder fields, and resolves both flags from builder values and environment variables during `HttpServiceConfigBuilder::build()` before constructing `State`.
Route registration gating `lib/llm/src/http/service/service_v2.rs`	Makes system routes mutable and conditionally registers the `busy_threshold` admin routes only when `admin_api_enabled` is true, logging when disabled.
Request handler nvext gating `lib/llm/src/http/service/openai.rs`	Applies `state.nvext_enabled()` check in `/v1/completions`, `/v1/chat/completions`, and `/v1/responses` handlers to conditionally allow or suppress client-supplied routing-override extensions.

🎯 2 (Simple) | ⏱️ ~12 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description provides comprehensive detail but does not include the required 'Related Issues' section, which is marked as required in the template.	Add a 'Related Issues' section with either 'Closes `#XXXX`' for linked issues or confirm no related issue exists with a checkbox.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically summarizes the main change: adding two master switches (DYN_ENABLE_NVEXT and DYN_ENABLE_FRONTEND_ADMIN_API) to the frontend service.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

coderabbitai

Actionable comments posted: 2

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

lib/runtime/src/config/environment_names.rs (1)
715-726: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Add new environment variable constants to the duplicate-check test.

The test array is missing the newly added constants llm::DYN_ENABLE_NVEXT and llm::DYN_ENABLE_FRONTEND_ADMIN_API. Without them, the test cannot detect duplicate names for these variables.
🔧 Suggested fix
             llm::DYN_ENABLE_ANTHROPIC_API,
+            llm::DYN_ENABLE_NVEXT,
+            llm::DYN_ENABLE_FRONTEND_ADMIN_API,
             llm::DYN_STRIP_ANTHROPIC_PREAMBLE,
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@lib/runtime/src/config/environment_names.rs` around lines 715 - 726, The
duplicate-check test's array of environment constants is missing the newly added
llm::DYN_ENABLE_NVEXT and llm::DYN_ENABLE_FRONTEND_ADMIN_API symbols; update the
test array in environment_names.rs (the list used for duplicate detection) to
include llm::DYN_ENABLE_NVEXT and llm::DYN_ENABLE_FRONTEND_ADMIN_API so the test
will detect duplicates for those new env vars.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@lib/llm/src/http/service/service_v2.rs`:
- Line 316: The doc comment in service_v2.rs currently references a removed
route (`POST /clear_kv_blocks`) alongside `busy_threshold`; remove the `POST
/clear_kv_blocks` mention from that doc comment so it only documents the
existing routes (e.g., leave `POST/GET /busy_threshold` and related text
intact). Locate the comment near the busy_threshold documentation in the
service_v2 module and update the comment text accordingly.

In `@lib/runtime/src/config/environment_names.rs`:
- Line 311: Remove the stale documentation bullet that references the removed
admin route `POST /clear_kv_blocks` in the doc comment inside
environment_names.rs (the comment block containing " - `POST /clear_kv_blocks`
(clear all KV cache blocks on workers)"). Edit that comment to delete the line
or update the list so it no longer lists `POST /clear_kv_blocks`, leaving other
documented routes unchanged.

---

Outside diff comments:
In `@lib/runtime/src/config/environment_names.rs`:
- Around line 715-726: The duplicate-check test's array of environment constants
is missing the newly added llm::DYN_ENABLE_NVEXT and
llm::DYN_ENABLE_FRONTEND_ADMIN_API symbols; update the test array in
environment_names.rs (the list used for duplicate detection) to include
llm::DYN_ENABLE_NVEXT and llm::DYN_ENABLE_FRONTEND_ADMIN_API so the test will
detect duplicates for those new env vars.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 8e0ea28f-48e9-4ae9-b9c1-f6759d631f26

📥 Commits

Reviewing files that changed from the base of the PR and between a95e0b1 and 9628a8a.

📒 Files selected for processing (3)

lib/llm/src/http/service/openai.rs
lib/llm/src/http/service/service_v2.rs
lib/runtime/src/config/environment_names.rs

- Drop POST /clear_kv_blocks from env-var docs (it's not a registered route) and delete the orphaned lib/llm/src/http/service/clear_kv_blocks.rs source file (dead since PR #1629 removed the route registration; no mod declaration, so it wasn't even compiled). - Rename State::new_with_flags -> State::new_with_nvext_enabled (only one flag for now). - Collapse the 6-line copy-pasted comment block at the 3 openai.rs handler call sites to a one-liner; rewrite the if/else as an expression assignment. - Trim const docs in environment_names.rs to behavior + default; the EPP/GAIE/Prime-RL compatibility paragraph is design rationale and lives in the PR description, not the public env-var reference. - Trim the env-mirror comment, the field/accessor doc duplication, and switch the admin-API-disabled log to structured fields. - Add two tests: enable_nvext propagates through the builder onto State.nvext_enabled (default + on + off), and enable_admin_api=false causes GET /busy_threshold to 404 while /live still serves 200. Signed-off-by: nnshah1 <neelays@nvidia.com>

Graham-review followup: the one-liner 'nvext master switch — see State::nvext_enabled' was copy-pasted at three handler call sites. state.nvext_enabled() is already self-documenting and the accessor's own /// doc points at the env var definition; the call-site comment adds nothing. Signed-off-by: nnshah1 <neelays@nvidia.com>

…, add env-var test, consolidate docs Addresses biswapanda's review feedback on the master-switch PR: 1. ("keep them consistent") Rename DYN_ENABLE_NVEXT to DYN_ENABLE_FRONTEND_NVEXT to match the DYN_ENABLE_FRONTEND_ADMIN_API sibling. Only env-var name changes; internal API (enable_nvext field, nvext_enabled() accessor) stays. 2. ("embeddings has no nvext_enabled() gate") KISS — instead of gating the embedding handler, remove nvext from the embedding protocol entirely. The embedding-local NvExt only carried an annotations field (SSE event triggers, useless for typically-non-streaming embeddings) and never carried any routing fields. Delete lib/llm/src/protocols/openai/embeddings/nvext.rs and drop the nvext: Option<NvExt> field from NvCreateEmbeddingRequest plus the AnnotationsProvider impl. preprocessor.rs's preprocess_embedding_request no longer reads request annotations. 3. ("add a test case for unset/true/false env-var with temp_env::with_vars") Add test_dyn_enable_frontend_nvext_env_var_mirror covering unset, truthy, falsey, and the builder-overrides-env case. 4. ("we'd update user-facing docs") Add a Frontend feature switches section under docs/components/frontend/configuration.md with one consolidated table covering DYN_ENABLE_FRONTEND_NVEXT and DYN_ENABLE_FRONTEND_ADMIN_API: default values, behavior when false, and compatibility notes. The existing /busy_threshold rows in the Infrastructure routes table cross-reference the new section. cargo check / cargo fmt --check / cargo clippy --tests --no-deps all clean. 4 unit tests pass: - test_liveness_endpoint_reflects_cancellation - test_enable_nvext_propagates_through_builder_to_state - test_dyn_enable_frontend_nvext_env_var_mirror (new) - test_admin_api_disabled_404s_busy_threshold Signed-off-by: nnshah1 <neelays@nvidia.com>

github-actions · 2026-06-11T06:32:39Z

🌿 Fern Docs Preview: https://nvidia-preview-444421ed-2135-4908-a46f-a5f1c67d6d81.docs.buildwithfern.com/dynamo/dev

@nnshah1

Reversal of the deletion in fd3a977. Per offline discussion with @nnshah1: even though the embedding-local NvExt currently only exposes 'annotations', deleting it removes a contract clients may be relying on. Keep the surface, gate its consumption — same pattern as the three other handlers — so DYN_ENABLE_FRONTEND_NVEXT=false closes off the *behavior* without breaking the request shape. Embedding handler now does the same one-line clear at handler entry: if !state.nvext_enabled() { request.nvext = None; } Routing-override headers don't apply to embeddings, so no apply_header_routing_overrides call to gate. Restored: - lib/llm/src/protocols/openai/embeddings/nvext.rs - lib/llm/src/protocols/openai/embeddings.rs (nvext field + NvExtProvider + AnnotationsProvider impls) - lib/llm/src/preprocessor.rs (request.has_annotation() + request.annotations() calls) Doc table updated to list /v1/embeddings alongside the other three gated handlers. Signed-off-by: nnshah1 <neelays@nvidia.com>

Add an end-to-end integration test asserting that when the frontend nvext extension is disabled, a request asking for response-side extra_fields (plus a routing header and body routing field) produces no nvext field in the response — proving the request- and response-side strips are wired together, not just unit-tested in isolation. Also warn once at the handler when a request carries nvext data (body or routing headers) while the extension is disabled, so the otherwise silent strip is observable in logs. The check lives only in the disabled branch, so the default request path is unaffected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: nnshah1 <neelays@nvidia.com>

Move warn_nvext_disabled above the handler_completions doc block (it had been inserted mid-doc-comment, splitting it) and trim verbose docstrings on the helper and the new test per review. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: nnshah1 <neelays@nvidia.com>

Replace the verbose master-switch preamble with a short deployment-focused note: extensions are on by default; disable the ones a deployment doesn't need to prevent accidental abuse. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: nnshah1 <neelays@nvidia.com>

harryskim

LGTM on doc

…I master switches (ai-dynamo#10556) Signed-off-by: nnshah1 <neelays@nvidia.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…-dynamo#6291) Operators need a runtime control to flush the reused KV prefix cache across every worker. The frontend had no such route, so the control returned HTTP 404. The earlier version of this PR added the route to clear_kv_blocks.rs. That module was removed upstream in ai-dynamo#10556, which reworked the frontend admin API behind the DYN_ENABLE_FRONTEND_ADMIN_API / DYN_DISABLE_FRONTEND_ADMIN_API master switch. This reimplements the route against the new admin API: - New module lib/llm/src/http/service/reset_prefix_cache.rs registers POST /reset_prefix_cache, following the busy_threshold admin-route pattern. The handler discovers every worker group exposing the clear_kv_blocks endpoint and fans the control out to each instance via PushRouter, reporting per-instance success/failure. - The route is registered only when the admin API is enabled, reusing the DistributedRuntime already plumbed through HttpServiceConfig.runtime. - Adds a #[tokio::test] in lib/llm/tests/http-service.rs asserting the route is mounted (returns 200 with a clear message rather than 404). Fixes ai-dynamo#6291 Signed-off-by: Asad Shahid <asad.shahid@berkeley.edu> Co-Authored-By: Claude <noreply@anthropic.com>

…-dynamo#6291) Operators need a runtime control to flush the reused KV prefix cache across every worker. The frontend had no such route, so the control returned HTTP 404. The earlier version of this PR added the route to clear_kv_blocks.rs. That module was removed upstream in ai-dynamo#10556, which reworked the frontend admin API behind the DYN_ENABLE_FRONTEND_ADMIN_API / DYN_DISABLE_FRONTEND_ADMIN_API master switch. This reimplements the route against the new admin API: - New module lib/llm/src/http/service/reset_prefix_cache.rs registers POST /reset_prefix_cache, following the busy_threshold admin-route pattern. The handler discovers every worker group exposing the clear_kv_blocks endpoint and fans the control out to each instance via PushRouter, reporting per-instance success/failure. - The route is registered only when the admin API is enabled, reusing the DistributedRuntime already plumbed through HttpServiceConfig.runtime. - Adds a #[tokio::test] in lib/llm/tests/http-service.rs asserting the route is mounted (returns 200 with a clear message rather than 404). Fixes ai-dynamo#6291 Signed-off-by: Asad Shahid <asad.shahid@berkeley.edu>

nnshah1 requested a review from a team June 10, 2026 20:28

pull-request-size Bot added the size/L label Jun 10, 2026

github-actions Bot added feat frontend `python -m dynamo.frontend` and `dynamo-run in=http|text|grpc` labels Jun 10, 2026

devin-ai-integration Bot reviewed Jun 10, 2026

View reviewed changes

Comment thread lib/runtime/src/config/environment_names.rs

coderabbitai Bot reviewed Jun 10, 2026

View reviewed changes

Comment thread lib/llm/src/http/service/service_v2.rs Outdated

Comment thread lib/runtime/src/config/environment_names.rs Outdated

copy-pr-bot Bot temporarily deployed to GITLAB June 10, 2026 22:37 Inactive

copy-pr-bot Bot temporarily deployed to GITLAB June 10, 2026 22:43 Inactive

nnshah1 enabled auto-merge (squash) June 10, 2026 23:23

copy-pr-bot Bot temporarily deployed to GITLAB June 10, 2026 23:48 Inactive

biswapanda reviewed Jun 11, 2026

View reviewed changes

Comment thread lib/runtime/src/config/environment_names.rs Outdated

biswapanda reviewed Jun 11, 2026

View reviewed changes

Comment thread lib/llm/src/http/service/service_v2.rs

biswapanda reviewed Jun 11, 2026

View reviewed changes

Comment thread lib/llm/src/http/service/openai.rs

biswapanda reviewed Jun 11, 2026

View reviewed changes

Comment thread lib/llm/src/http/service/service_v2.rs

nnshah1 requested a review from a team as a code owner June 11, 2026 06:29

pull-request-size Bot added size/XL and removed size/L labels Jun 11, 2026

copy-pr-bot Bot temporarily deployed to GITLAB June 11, 2026 06:30 Inactive

github-actions Bot added the documentation Improvements or additions to documentation label Jun 11, 2026

copy-pr-bot Bot temporarily deployed to GITLAB June 11, 2026 06:30 Inactive

biswapanda approved these changes Jun 11, 2026

View reviewed changes

pull-request-size Bot added size/L and removed size/XL labels Jun 11, 2026

copy-pr-bot Bot temporarily deployed to GITLAB June 11, 2026 06:53 Inactive

copy-pr-bot Bot temporarily deployed to GITLAB June 11, 2026 17:53 Inactive

pull-request-size Bot added size/XL and removed size/L labels Jun 11, 2026

copy-pr-bot Bot temporarily deployed to GITLAB June 11, 2026 18:32 Inactive

copy-pr-bot Bot temporarily deployed to GITLAB June 11, 2026 18:42 Inactive

harryskim approved these changes Jun 11, 2026

View reviewed changes

nnshah1 merged commit 07d9806 into main Jun 11, 2026
93 checks passed

nnshah1 deleted the neelays/dyn-enable-nvext-and-admin-api branch June 11, 2026 19:27

This was referenced Jun 30, 2026

refactor(frontend): rename DYN_ENABLE_FRONTEND_{NVEXT,ADMIN_API} to DYN_DISABLE_* #11101

Merged

[cherry-pick 1.3.0] refactor(frontend): rename DYN_ENABLE_FRONTEND_{NVEXT,ADMIN_API} to DYN_DISABLE_* #11123

Merged

glamr-agent mentioned this pull request Jul 29, 2026

fix(codeowners): repoint orphaned planner+operator glob at moved kubernetes_api.py #12378

Merged

Navigation Menu

Uh oh!

feat(frontend): add DYN_ENABLE_NVEXT and DYN_ENABLE_FRONTEND_ADMIN_API master switches - #10556

feat(frontend): add DYN_ENABLE_NVEXT and DYN_ENABLE_FRONTEND_ADMIN_API master switches#10556
nnshah1 merged 8 commits into
mainfrom
neelays/dyn-enable-nvext-and-admin-api

nnshah1 commented Jun 10, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

Uh oh!

coderabbitai Bot commented Jun 10, 2026

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

github-actions Bot commented Jun 11, 2026 •

edited

Loading

Uh oh!

harryskim left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

nnshah1 commented Jun 10, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Motivation

Compatibility

Design notes

Test plan

Out of scope (follow-ups)

Summary by CodeRabbit

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

coderabbitai Bot commented Jun 10, 2026

Walkthrough

Changes

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

github-actions Bot commented Jun 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

harryskim left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

nnshah1 commented Jun 10, 2026 •

edited by coderabbitai Bot

Loading

github-actions Bot commented Jun 11, 2026 •

edited

Loading