github maximhq/bifrost ent-v2.1.0-base
Enterprise v2.1.0 base version
on GitHub

2 hours ago

What's Changed

  • feat: add fallback UI for vk creation page for AP-managed VKs by @roroghost17 in #6618
  • docs: update AP constraints for user created vks by @roroghost17 in #6619
  • fix(ui): make use_deployments_endpoint optional, improve Replicate deployment descriptions, and add form error logging by @impoiler in #6596
  • feat(ui): display canonical model name with fallback to model in logs columns by @impoiler in #6602
  • feat(ui): add RawJsonUnavailableNotice explaining disabled raw storage in log detail Raw JSON tab by @impoiler in #6608
  • fix(ui): use server-computed flag as sole source of truth for managed VK state and add tooltip to disabled active switch by @impoiler in #6610
  • fix(configstore): preserve created_at on routing rule update and sync by @impoiler in #6616
  • fix: realtime / ws responses adds llm.call spans by @roroghost17 in #6592
  • fix: routing-rule reads honor the row-visibility queryscope by @akshaydeo in #6638
  • feat: add rotation grace-period columns to governance_virtual_keys by @akshaydeo in #5437
  • feat: add vk_rotation_cooldown client config setting by @akshaydeo in #5438
  • feat: honor rotation grace-period previous value in governance auth by @akshaydeo in #5439
  • feat: apply rotation cooldown in the virtual key rotate endpoints by @akshaydeo in #5440
  • feat: reconcile config.json virtual key values against rotation state by @akshaydeo in #5441
  • feat: virtual key rotation cooldown UI and OpenAPI docs by @akshaydeo in #5442
  • updated sheet design by @akshaydeo in #6651
  • docs fixes by @akshaydeo in #6677
  • feat: add grant implementation for per-request identity, access, and limits by @Pratham-Mishra04 in #6306
  • test: add unit tests for EffectiveAccess, Grant, context propagation, and uniqueEntries in the grants package by @Pratham-Mishra04 in #6307
  • feat: settle request identity onto a grant at context conversion, recording credential and user from headers, middleware context values, MCP JWT, ephemeral keys, and WebSocket upgrade by @Pratham-Mishra04 in #6641
  • feat: wire up grant creation in governance by @Pratham-Mishra04 in #6308
  • refactor: replace VK-specific provider filter with ResolveEffectiveAccess on model listing routes so governance-resolved access narrows the fan-out instead of raw DB rows by @Pratham-Mishra04 in #6309
  • feat: add split-commit-into-stack Claude skill for analyzing, partitioning, and proving stacked Graphite PRs by @Pratham-Mishra04 in #6310
  • refactor: unify governance limit assembly, enforcement, and billing onto a single resolved access so checked, recorded, and charged sets cannot diverge by @Pratham-Mishra04 in #6311
  • refactor: replace fieldset[disabled/inert] managed-profile guard with conditional render and update access-profile alert copy by @Pratham-Mishra04 in #6312
  • refactor: replace *TableVirtualKey in routing rule evaluation with a GovernanceScope struct read off the resolved access context, eliminating a duplicate credential walk by @Pratham-Mishra04 in #6313
  • feat: add GetMCPClientNames to InMemoryStore for full client ID→name mapping across governance and config layers by @Pratham-Mishra04 in #6314
  • feat: rename AllowOnAllVirtualKeys to AllowByDefault across schema, store, API, and UI, with backward-compatible wire aliases for both JSON keys and query params by @Pratham-Mishra04 in #6642
  • refactor: replace per-VK MCP server map with a single atomic server, moving tool-access scoping from server selection to governance admission via MCPGatewayAdmitter, and unify /mcp auth into authenticate/admit with status-mapped refusals by @Pratham-Mishra04 in #6643
  • fix: mark internal MCP health-check/admin-verify contexts with BifrostContextKeyMCPHealthCheckRequest, thread grant through foreign context wrappers via bifrostContextSelfKey, suppress governance warn→debug for grantless probes, hide profile-owned VK by @Pratham-Mishra04 in #6649
  • fix: drop legacy FK constraints on oauth_user_tokens/oauth_user_sessions blocking DeleteMCPClientConfig after token-table merge, with -:migration struct tags and regression tests by @Pratham-Mishra04 in #6648
  • chore: rename virtual_key_not_found/virtual_key_blocked error codes to access_not_found/access_blocked in e2e test assertions by @Pratham-Mishra04 in #6678
  • perf: add focused usage snapshot by @danpiths in #6683
  • community: add Magic Hour to MCP library by @runshouse in #6691
  • fix: propagate grant from submit context into async job's background context via explicit SetGrant call by @Pratham-Mishra04 in #6701
  • fix: revert ResolveAccess side-effect in PreMCPConnectionHook with direct VK lookup to populate identity context keys by @Pratham-Mishra04 in #6706
  • codeql fixes by @akshaydeo in #6696
  • feat(logstore): add request_id exact PK lookup filter, bypassing time-range constraints by @impoiler in #6694
  • feat(ui): add request ID search mode to logs with UUID auto-detection, id: prefix support, and ID badge by @impoiler in #6695
  • refactor(ui): extract shared filter primitives and webhook delivery utils into dedicated modules by @impoiler in #6707
  • feat(logstore): add cross-endpoint GET /api/webhooks/deliveries search with filterable delivery history by @impoiler in #6708
  • feat(ui): add webhook deliveries page with filtering, pagination, and redelivery by @impoiler in #6709
  • feat(ui): add fleet-wide webhook deliveries page with cross-endpoint filtering and replace sheet pagination with preview link by @impoiler in #6710
  • feat(ui): replace in-page back button on webhook deliveries with topbar breadcrumb trail by @impoiler in #6714
  • feat(logs): add compare_to_previous to GET /api/logs/stats by @akshaydeo in #6719
  • feat(logs): replace the metric cards with the segmented strip by @akshaydeo in #6720
  • [feat]: add Baseten to Hugging Face inference providers by @nicolastoulemont in #6633
  • genai signature drop fix by @akshaydeo in #6745
  • [fix]: Accept model arrays from OpenAI-compatible APIs by @dani29 in #6712
  • [fix]: marshal required nullable response fields by @PSR94 in #6723
  • fix: pass mime type in file responses by @TejasGhatte in #6684
  • fix: send rerank documents as Cohere v2 strings, not objects by @AdityaPainuli in #6654
  • fix: output resolution costing for video requests by @TejasGhatte in #6672
  • fix: jobaccounting package migration by @TejasGhatte in #6673
  • fix: update table batch job to provider job by @TejasGhatte in #6674
  • feat: video costs wired in job accounting by @TejasGhatte in #6675
  • fix: video debug in logs and retry fix in polling by @TejasGhatte in #6728
  • mcp auth with ssrf checks by @akshaydeo in #6757
  • tests: files api harness tests by @TejasGhatte in #6765
  • require auth for realtime connections when enforced on inference by @akshaydeo in #6759
  • block Teredo addresses in the dial-time SSRF guard by @akshaydeo in #6760
  • scope prompt child reads and writes to their parent prompt by @akshaydeo in #6761
  • fix: bedrock models routing to converse by @TejasGhatte in #6655
  • enterprise release notes by @akshaydeo in #6771
  • feat(anthropic): synthesize missing terminal text suffix from output_text.done by @Madhuvod in #6159
  • feat(integrations): support native raw request redaction by @Madhuvod in #6365
  • feat(streaming): add RawStreamTextCodec and paused buffer transform for native SSE redaction by @Madhuvod in #6386
  • helm+docs changes for the send_all_conversation_turns setting in guardrail rules by @Madhuvod in #6758
  • feat: replace user-scoped model configs with extensible ExtraScopedIDsResolver and add ManagedBy field, read-only sheet, and managed_by API response by @BearTS in #6715
  • feat: add ReadOnlyNotice component to model limit scope registry for customizable read-only alert messaging by @BearTS in #6733
  • feat(quota): tag budgets/rate limits with their source, list every rate limit by @BearTS in #6752
  • [fix]: ui - declare VKCreationPolicyResponse in the OSS access-profile fallback by @markdawson in #6776
  • fix: remove score filter from redis by @TejasGhatte in #6772
  • fix: hex value fields in redis store by @TejasGhatte in #6773
  • fix: apply rate limits on model less passthrough requests by @TejasGhatte in #6774
  • feat(ui): audit virtual key reveal and copy events via enterprise hook with OSS no-op fallback by @impoiler in #6786
  • fix(ui): replace hard-coded metric strip colours with theme tokens and stabilise strip height across loading states by @impoiler in #6788
  • fix(ui): fix logs page layout on small screens with scrolling fallback and header wrap fixes by @impoiler in #6789
  • feat: route azure deepseek responses requests to chat completions for coding harnesses and fall back for models lacking responses endpoint support for bedrock by @sammaji in #6326
  • fix(ui): move sheet header padding from container to sticky header element across all sheets by @impoiler in #6790
  • feat: add compat toggle to gate chat completions conversion for azure deepseek models by @sammaji in #6737
  • fix: use max_tokens instead of max_completion tokens for ollama by @sammaji in #6607
  • feat: add structured logging for compat plugin param drops and conversions by @sammaji in #6634
  • fix: route Anthropic-on-Vertex passthrough usage and stream terminal detection through Anthropic parsers by @sammaji in #6639
  • chore: adds vertex / azure / bedrock test cases to harness by @sammaji in #6738
  • fix(configstore): preserve created_at across config sync for budget, rate limit, team, customer, model config, pricing override, and plugin by @impoiler in #6792
  • fix(ui): pause background polling while edit sheets are open to prevent in-progress edits from being discarded by refetches by @impoiler in #6793
  • fix: remove raw body passthrough if provider does not support output config format by @TejasGhatte in #6798
  • feat: add SecretVar config support and ConfigMarshallerPlugin interface to native .so plugins by @BearTS in #6600
  • feat: add project dimension to governance context, log tables, filters, and permit evaluation with refusal when a named project goes unscoped by @Pratham-Mishra04 in #6702
  • feat: add Projects governance page with RBAC gating and sidebar entry by @Pratham-Mishra04 in #6703
  • feat: add project as a first-class filter, histogram, and ranking dimension across API, matview, and UI by @Pratham-Mishra04 in #6704
  • feat: add project_id/project_name governance dimensions to span attributes, enrichment dims, metrics labels, and log payloads by @Pratham-Mishra04 in #6705
  • refactor: move gatherLimits into GovernanceStore.GatherLimits, propagate settling errors as DecisionAccessBlocked after access checks, and split modelConfigScopesFor to one permit per call by @Pratham-Mishra04 in #6724
  • refactor: replace []string fields on ProviderPermit/MCPPermit/ProviderCandidate with schemas.WhiteList/schemas.BlackList types, moving list semantics onto the types and deleting lists.go by @Pratham-Mishra04 in #6725
  • removes heartbeats from genai integration by @akshaydeo in #6804
  • refactor: split ProviderAndModelLimits into GlobalProviderLimits, GlobalModelLimits, PermitProviderLimits, and PermitModelLimits, adding ProviderScopedModelLimits for load-balancing candidate exclusion by @Pratham-Mishra04 in #6729
  • feat: add governance.projects to config.json schema and docs with reconciliation semantics, budget/rate-limit fields, and enterprise-only exclusion in OSS tests by @Pratham-Mishra04 in #6777
  • feat: add credential block to MCPClientResponse with scopes and refresh-token presence, surface in server sheet and sessions table by @Pratham-Mishra04 in #6780
  • fix: remove plugin versioning field and adds hash-based merge logic by @sammaji in #6250
  • chore: updates docs to remove plugin versioning by @sammaji in #6336
  • chore: updates helm schema to remove plugin versioning by @sammaji in #6337
  • feat: persist and display served model from provider response by @sammaji in #6693
  • feat: record MCPConnectionFailure (stage, message, timestamps) on client state and credential rows, surfacing failure reason in client-list API and cluster node states by @Pratham-Mishra04 in #6794
  • feat: surface MCP connection failure details in state badge popover and server sheet, with shared mcpConnectionFailure utils, credential status_reason, and unified credential/session status badge palette by @Pratham-Mishra04 in #6795
  • docs: add last_failure/node_states fields to MCPClient schema and document connection-state failure reasons by @Pratham-Mishra04 in #6796
  • fix: propagate request grant into relay context and export SettleIdentity for WebSocket upgrade path so governance sees a settled identity on every turn by @Pratham-Mishra04 in #6801
  • feat(governance): add ScopedModelLimits for named-scope model limit lookup without a permit by @BearTS in #6800
  • fix: resolve mcp client id before vault hooks are called via gorm by @BearTS in #6812
  • fix: initialize new model budgets with empty array instead of default budget entry by @BearTS in #6813
  • feat: add ScopedModelConfigs cache accessor and preserve budget id through model-budget edits in ProviderConfigCard by @Pratham-Mishra04 in #6810
  • feat: add log-level filtering to plugin and routing decision logs with shared logLevel utils, level badges, and LogLevelTabs component by @Pratham-Mishra04 in #6811
  • fix: anthropic stream truncation by @TejasGhatte in #6805
  • feat: add github-copilot provider by @akshaydeo in #6352
  • feat(docs): add new providers (bedrock_mantle, opencode-go, opencode-zen, runware, sarvam, wafer) and alphabetize provider enums by @impoiler in #6664
  • refactor: remove counts prop and countAtOrAboveEachLevel from LogLevelTabs, routing logs, and plugin logs by @Pratham-Mishra04 in #6814
  • [fix]: Bedrock provider - placeholder content block instead of null on empty assistant messages by @Jesse-Schultz-Relativity in #6732
  • feat(providers): add Databricks provider with Model Serving and Unity AI Gateway support, PAT and OAuth M2M auth, and Gemini stream-options fixup by @impoiler in #6665
  • feat(configstore): add Databricks key config persistence with DB columns, encryption, redaction, validation, and merge support by @impoiler in #6666
  • feat(schema): add Databricks provider support with workspace URL, OAuth M2M, and AI Gateway routing config by @impoiler in #6667
  • feat(ui): add Databricks provider with PAT and OAuth M2M auth, inference surface selector, and governance tag forwarding by @impoiler in #6668
  • feat(providers): add Databricks as a supported provider with pricing catalog support and migration CLI exclusion by @impoiler in #6669
  • feat(docs): add Databricks as a first-class provider with PAT and OAuth M2M auth, Model Serving and Unity AI Gateway surfaces, and config.json examples by @impoiler in #6670
  • test(databricks): add provider harness coverage and CI secrets by @impoiler in #6671
  • fix(databricks): surface the upstream error message on inference failures by @impoiler in #6676
  • fix(ui): fix Databricks tab layout collapsing and OAuth M2M key validation after form reset by @impoiler in #6679
  • feat(databricks): gate optional parameters on datasheet capabilities per model and drop Anthropic-native fields unconditionally by @impoiler in #6770
  • feat(routing): three-tier complexity contracts, semantic config wire, and migrations by @kohlivrinda in #6163
  • feat(ui): remove provider governance table from model provider config view by @impoiler in #6820
  • feat(routing): vector stores, chromem backend, and the semantic embedding executor by @kohlivrinda in #6164
  • feat(routing): semantic classifier runtime and harness-aware input extraction by @kohlivrinda in #6165
  • feat(routing): semantic classification on the request path, with logs, traces, and telemetry by @kohlivrinda in #6166
  • feat(routing): semantic router API surface and configuration UI by @kohlivrinda in #6167
  • docs(routing): semantic complexity router helm values, OpenAPI, and documentation by @kohlivrinda in #6168
  • fix(logstore): repair racy has_object waits and stale payload count in tests by @kohlivrinda in #6176
  • feat(routing): LLM classifier fallback for complexity routing by @kohlivrinda in #6177
  • docs(complexity): OpenAPI, Helm, and feature doc coverage for the LLM fallback classifier by @Madhuvod in #6282
  • feat(complexity): session aware routing changes by @Madhuvod in #6317
  • renaming cachedebug -> cachemetadata + guardrail, routingmetadaata at all places by @Madhuvod in #6722
  • feat(complexity): enforce 750-phrase combined semantic limit with split-config merge guard and local Chromem cleanup fixes by @Madhuvod in #6727
  • feat complexity: warm coordination and generation reclamation by @kohlivrinda in #6807
  • fixes race conditions while closing the the streaming calls (fasthttp patch) by @akshaydeo in #6799
  • release connections held by abandoned streams by @akshaydeo in #6802
  • probe pool health after a cancelled stream in the harness by @akshaydeo in #6803
  • feat: mint github-copilot tokens server-to-server by @akshaydeo in #6353
  • feat: persist github-copilot app credentials by @akshaydeo in #6354
  • feat: github-copilot provider form and constants by @akshaydeo in #6355
  • test: wire github-copilot live scenarios and CI by @akshaydeo in #6356
  • docs: github copilot provider by @akshaydeo in #6357
  • feat: add virtual mcps servers compatibility by @roroghost17 in #6746
  • fix: connectors test fixes by @roroghost17 in #6635
  • feat: add vMCP CRUD support by @roroghost17 in #6747
  • feat: add all converter's internal processing to individual span buckets by @roroghost17 in #6588
  • docs: add overhead latency breakdown explanation page by @roroghost17 in #6637
  • feat: add support for serving virtual MCPs at slug endpoints by @roroghost17 in #6748
  • refactor: promote the MCP tool accumulator to a reusable component by @roroghost17 in #6749
  • feat: add vk assignments list support to vmcps and fix whitelisting convention for tools by @roroghost17 in #6750
  • feat: add virtual mcp UI by @roroghost17 in #6751
  • feat: wire in endpoint support in direct MCP servers by @roroghost17 in #6791
  • feat: add VMCP assignment into VK sheets by @roroghost17 in #6826
  • feat: add Allow all providers option on vks by @roroghost17 in #6662
  • feat: add allow_all_providers support in helm by @roroghost17 in #6663
  • docs: add Allow all providers docs by @roroghost17 in #6827
  • dependabot updates by @akshaydeo in #6832
  • feat: add multi-scope filter support for model configs with filterUnderScope registry field and comma-separated scope query param by @BearTS in #6829
  • feat: add scopes filter in model config api by @BearTS in #6830
  • faker patches for dependabot by @akshaydeo in #6835
  • codeql fixes by @akshaydeo in #6836
  • fix: openai reasoning content in bedrock converse response by @TejasGhatte in #6834
  • pending fixes from dependabot by @akshaydeo in #6837
  • feat(ui): prompt users to switch to first-party integration when custom provider name collides with a natively supported provider by @impoiler in #6845
  • feat: add ProviderScopedModelLimitsInScope for named-scope provider-scoped model limits in load balancing by @Pratham-Mishra04 in #6856
  • refactor: extract NarrowListModelsProviders onto BifrostHTTPServer and thread AccessResolver into integration routers so all models listings share one fan-out narrowing rule by @Pratham-Mishra04 in #6857
  • feat: add TruncatedBadge component and footer prop to EntitySelector by @BearTS in #6843
  • feat: replace managed_by/Source strings with structured SourceRef embedding across model configs and quota budgets by @BearTS in #6858
  • fix: virtual key loading state and consistency by @roroghost17 in #6859
  • fix(complexity): remove session xml tag check in cc and few complexity ui fixes by @Madhuvod in #6838
  • complexity fixes + added session id in logs table so its filtering on llm logs by @Madhuvod in #6846
  • feat: helm version release 2.1.38 with validation fix for postgresql.external.passwordCommand by @BearTS in #6861
  • feat: add VirtualKeyPruneGuard to protect access-profile virtual keys from config.json reconciliation pruning by @BearTS in #6863
  • complexity docs changes + openapi fix by @Madhuvod in #6865
  • fix: restore explicit opt-in for Infisical secrets by @danpiths in #6866
  • fix: encrypt rows one-per-transaction with cursor pagination and concurrent vault writes to prevent deadlocks and boot hangs by @sammaji in #6808
  • chore: adds ui tooltip to show cached and uncached input tokens breakdown by @sammaji in #6833
  • test: video accounting tests by @TejasGhatte in #6839
  • fix: helm chart to render the scim config as it is rather than not at all if enabled: false by @BearTS in #6869
  • fix(ui): prevent logout 401 cascade by latching logout state, resetting API cache after navigation, and stopping WebSocket reconnection on sign-out by @impoiler in #6855
  • feat(ui): add scope labels, maxVisible overflow, and scopeLabel prop to budget and rate limit displays by @impoiler in #6860
  • fix: don't serve disabled MCP clients or stale virtual MCP defs by @roroghost17 in #6872
  • docs: add virtual mcp docs by @roroghost17 in #6873
  • docs: refactor mcp gateway docs to overview by @roroghost17 in #6874
  • docs: add allow all providers changes by @roroghost17 in #6875
  • feat(ui): add guided migration dialog for custom Databricks providers to the first-party Databricks provider by @impoiler in #6876
  • handle azure providers for reasoning efforts by @akshaydeo in #6877
  • chore: replace FolderKanban with SquareKanban icon and fix sidebar className indentation by @Pratham-Mishra04 in #6878
  • fix: skip custom provider in list models call if request not allowed by @TejasGhatte in #6853
  • fix: thinking block modification error by @TejasGhatte in #6854
  • [fix]: preserve GPT-6 Astra max reasoning effort by @nettee in #6881
  • fix: add min-w-0 to SCIM page wrapper div by @BearTS in #6871
  • fix: map Anthropic forced tool choice any to required on OpenAI egress by @Atharva-Kanherkar in #6888
  • feat: add onSelect callback support to breadcrumb for non-URL parent views by @BearTS in #6893
  • openrouter prompt cache fixes by @akshaydeo in #6692
  • feat(core): synthesize prompt-cache breakpoints for clients that send none by @akshaydeo in #6697
  • feat(logstore): add request-type visibility filtering by @impoiler in #6890
  • test: cover the prompt-cache dispatch seam and reconcile the never-invent comment by @akshaydeo in #6698
  • docs: add Projects enterprise feature with full configuration, access rules, accounting modes, and observability attribution by @Pratham-Mishra04 in #6896
  • feat(dashboard): hide configured request types from log reads by @impoiler in #6891
  • feat: extend prompt-cache breakpoint translation to the gpt-5.6 family by @akshaydeo in #6699
  • feat(helm): expose hidden request types and document configuration by @impoiler in #6892
  • docs: add virtual_mcps to project schema, note API membership, and tighten wording from "grants" to "allows" by @Pratham-Mishra04 in #6897
  • docs: document auto-injected prompt cache breakpoints by @akshaydeo in #6700
  • feat(ui): display hidden request types in logging config view by @impoiler in #6894
  • test: cover the whole prompt-cache injection config surface in the harness by @akshaydeo in #6753
  • feat(ui): hide unreleased github-copilot provider from the UI while preserving config.json rendering by @impoiler in #6895
  • feat: document the Projects management API and project surface across the governance and logging specs by @Pratham-Mishra04 in #6898
  • docs: add throughput histograms, user rankings, and /inspect endpoint to OpenAPI spec and docs nav by @Pratham-Mishra04 in #6899
  • conflict fix by @akshaydeo in #6909
  • second conflict by @akshaydeo in #6910
  • docs: replace business-unit team membership with user membership (M:N) and add user VK creation policy endpoint by @BearTS in #6906
  • docs: add roles, business units, access profile mappings, and multi-profile docs across enterprise guides by @BearTS in #6907
  • feat(logstore): add tool_call_names column and filter by @impoiler in #6911
  • vk allowed models * handling for governance by @akshaydeo in #6767
  • fix(logging): populate tool_calls and tool_call_names on log entries by @impoiler in #6912
  • test(governance): cover MCP connect identity stamping by @akshaydeo in #6768
  • feat(logs): add tool call names filter to the logs API and UI by @impoiler in #6913
  • fix(logstore): make filterdata search case-insensitive across all backends by @impoiler in #6915
  • fix: forced tool choice capability flag by @TejasGhatte in #6903
  • feat: add scheduled automatic virtual key rotation by @impoiler in #6806
  • fix: delete stale routing rules in merge transaction before inserts to avoid priority collisions by @BearTS in #6934
  • feat: config json and helm updates for virtual MCPs by @roroghost17 in #6904
  • docs: update docs for virtual mcps config json changes by @roroghost17 in #6905
  • fix: preserve budget state across edits by @danpiths in #6932
  • chore: add vMCP e2e tests by @roroghost17 in #6919
  • feat: wire in tracing for async jobs by @roroghost17 in #6939
  • fix(ui): misc layout and styling fixes for providers, sheets, and classifier status badge by @impoiler in #6940
  • feat: helm changes for projects, Databricks, access-profile mappings, and vk rotation cooldown by @BearTS in #6917
  • fix: derive chat finish_reason from responses terminal state and incomplete details by @sammaji in #6901
  • fix: populate summary_index, summary text, and signatures on reasoning_summary_* stream events for anthropic, bedrock, and gemini by @sammaji in #6902
  • fix: normalize vertex genai resource model names to bare ids for governance and key selection by @sammaji in #6918
  • fix: preserve plugin created_at timestamp on update by @sammaji in #6935
  • chore: test fixes and updates provider harness by @sammaji in #6920
  • fix: preserve realtime auth across KV replication by @danpiths in #6943
  • chore: test fixes by @sammaji in #6938
  • docs: add description for scope query parameter in model-configs endpoint by @BearTS in #6946
  • docs: remove redundant info callout about user-focused scope in access-profiles by @BearTS in #6947
  • fix: rename assistant reasoning_content to alias for groq and cerebras instead of stripping by @sammaji in #6949
  • chore: update test models for fireworks, groq, huggingface, mistral, and vertex by @sammaji in #6950
  • feat: Refactor provider configs editor to shared component by @roroghost17 in #6951
  • feat: add allow all providers for projects in config json by @roroghost17 in #6952
  • chore: update helm values for allow all providers by @roroghost17 in #6953
  • fix: add custom provider toggle to differentiate in providers who send done marker by @TejasGhatte in #6948
  • tests: add e2e tests for allow all providers on vk by @roroghost17 in #6954
  • fix: ui fixes for mcp and vmcp by @roroghost17 in #6957
  • feat: add toggle for overhead spans filter in plugins by @roroghost17 in #6945
  • feat(databricks): add Claude reasoning translation, image inlining, and Responses API fallback by @impoiler in #6958
  • fix: terminate stream for upstream providers who sends heart beats after usage by @TejasGhatte in #6960
  • fix: bedrock dropping reasoning sign by @akshaydeo in #6942
  • docs: document API-key auth for prometheus /metrics endpoint by @roroghost17 in #6963
  • feat(config): move hidden_request_types to client config with UI editing support by @impoiler in #6962
  • feat(databricks): auto-prefix bare model names with system.ai. for the Unity AI Gateway by @impoiler in #6965
  • feat(databricks): add model name refinement with Unity Catalog prefix and endpoint passthrough by @impoiler in #6968
  • harness fixes by @akshaydeo in #6969
  • azure streaming fallback fixes by @akshaydeo in #6975
  • fix: add atomic signed budget usage updates by @danpiths in #6976
  • v2.1.0 by @akshaydeo in #6979
  • helm check fixes by @akshaydeo in #6982
  • chore: bump bifrost helm chart version to 2.1.39 by @BearTS in #6981
  • helm test fixes and helm version bump by @akshaydeo in #6983
  • fix: add project and complexity fields to dimensions list by @roroghost17 in #6985
  • test harness fixes by @akshaydeo in #6986
  • chore: test fixes for authenticated MCPs by @roroghost17 in #6991
  • test harness fixes by @akshaydeo in #6992
  • routing plugin updates by @akshaydeo in #6994
  • plugin dependencies update by @akshaydeo in #6995
  • fixes version by @akshaydeo in #6996
  • fixes version (#6996) by @akshaydeo in #6997

New Contributors

Full Changelog: ent-v2.0.1-base...ent-v2.1.0-base

Check out latest releases or
releases around maximhq/bifrost ent-v2.1.0-base

Don't miss a new bifrost release

NewReleases is sending notifications on new releases.

Get notifications