-/https://github.com/berriai/litellm/issues/28585

8 views

-/https://github.com/berriai/litellm/issues/28585

GitHub · issue

#28585 Agents page can hit 422 from /key/list page size limit

State: open
Author: @ogulcansarioglu

## Summary

The Agents dashboard can fail while fetching key metadata because the UI calls `/key/list` with a page size larger than the backend accepts.

## Symptom

A downstream deployment observed the Agents page request failing with `422 Unprocessable Content` for:

`GET /key/list?page=1&size=500&return_full_object=true&include_team_keys=true&include_created_by_keys=true`

The backend route validates `size` with a maximum of `100`, so `size=500` is rejected before the key-list response can be returned.

## Fix PR

PR: https://github.com/BerriAI/litellm/pull/28578

The PR: - changes the dashboard key lookup page size to `100` - paginates `/key/list` until all visible agent keys are found or all pages are exhausted - adds regression coverage for the page size and pagination behavior - keeps dependencies and lockfiles unchanged

Current PR status at the time of filing: - GitHub Actions pass - Veria AI - PR Review passes - PR is mergeable - local `npm audit signatures` passed

cc @krrish-berri-2 because recent Agents dashboard work points to that area of ownership.

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

vote history (7 events)

#0 of 0 · 31d19h14m1s ago — entered · #import:https:::github.com:berriai:litellm post #1640

-/https://github.com/berriai/litellm/issues/304683:1-/https://github.com/berriai/litellm/issues/28585⭐

The right item is harder because it spans shared UI rendering and copy-generation paths, with added header serialization and sensitive-data handling concerns. The left has a focused, already-implemented change with regression coverage.

#0 of 0 · 31d19h12m28s ago · #import:https:::github.com:berriai:litellm post #1667

-/https://github.com/berriai/litellm/issues/3505610:1-/https://github.com/berriai/litellm/issues/28585⭐

The left task has broader cross-layer scope, involving provider integration, multiple data modalities, streaming or binary handling, compatibility behavior, tests, and documentation. The right task is a localized dashboard pagination and validation adjustment with existing implementation guidance, so it carries substantially less engineering risk.

#0 of 0 · 31d19h12m16s ago · #import:https:::github.com:berriai:litellm post #1671

-/https://github.com/berriai/litellm/issues/296618:1-/https://github.com/berriai/litellm/issues/28585⭐

The left requires cross-cutting backend authentication work involving cloud identity, token lifecycle, configuration compatibility, database reconnect behavior, and security-focused testing. The right is a contained dashboard/API pagination correction with an existing implementation path and regression coverage, so the left carries substantially greater engineering risk and scope.

#0 of 0 · 31d19h11m59s ago · #import:https:::github.com:berriai:litellm post #1676

-/https://github.com/berriai/litellm/issues/28585⭐3:1-/https://github.com/berriai/litellm/issues/26170

Issue 28585 is harder because it spans dashboard request behavior, pagination control flow, API-limit compatibility, and regression coverage; issue 26170 is a localized precedence correction across three provider paths with comparatively contained testing.

#0 of 0 · 31d19h8m16s ago · #import:https:::github.com:berriai:litellm post #1739

-/https://github.com/berriai/litellm/issues/347104:1-/https://github.com/berriai/litellm/issues/28585⭐

The left requires security-sensitive data-flow auditing, robust sanitization across multiple logging construction paths, regression coverage, and careful compatibility validation. The right is a localized dashboard pagination and request-limit adjustment with an outlined implementation, making it substantially lower risk and scope.

#0 of 0 · 31d18h21m43s ago · #import:https:::github.com:berriai:litellm post #2488

-/https://github.com/berriai/litellm/issues/268063:2-/https://github.com/berriai/litellm/issues/28585⭐

Issue #26806 is harder because it requires tracing fallback model metadata through pricing and cache-billing logic, preserving several related cost variants, and adding backend regression coverage across custom-model scenarios. Issue #28585 is a more localized dashboard request/pagination correction with a clearer validation boundary.

#0 of 0 · 31d18h18m37s ago — current · #import:https:::github.com:berriai:litellm post #2536

-/https://github.com/berriai/litellm/issues/339558:1-/https://github.com/berriai/litellm/issues/28585⭐

The left issue requires a cross-cutting backend data-path redesign involving bounded memory, streaming or chunked processing, worker-concurrency behavior, cleanup, and regression testing under large payloads. The right issue is a localized dashboard/API pagination adjustment with a clear existing fix path and limited validation scope.

discussed in #import:https:::github.com:berriai:litellm

ranked child groups
scope depth

no voted pairs yet in this scope

cli

spread