Enterprise v1.5.3 base

What's Changed

• chore: build fix ui by @TejasGhatte in #4827
• feat: lower auth_code_ttl default to 300s and enforce 900s maximum by @Pratham-Mishra04 in #4822
• enterprise changelog by @akshaydeo in #4837
• fix(anthropic): preserve file id document sources by @mmacvicar in #4832
• fix(gemini): preserve file upload mime type by @mmacvicar in #4833
• fix: test fixes by @roroghost17 in #4823
• feat: add audit logs public API docs by @impoiler in #4847
• feat: add connection_type, auth_type, state, and virtual_key filters to MCP clients list by @impoiler in #4839
• feat: add filter sidebar with faceted filtering to MCP clients page by @impoiler in #4840
• feat: add pagination and filtering to MCP clients list endpoint by @impoiler in #4841
• fix: skip background token refresh for disabled/unconfigured MCP clients and guarantee non-nil logger in sync workers by @Pratham-Mishra04 in #4848
• docs: clarify two-layer token refresh behavior and disabled-client refresh token expiry by @Pratham-Mishra04 in #4849
• fix: perplexity responses api compatibility by @TejasGhatte in #4813
• fix: cancelled state in logs by @TejasGhatte in #4831
• fix: gemini openai through signature compatibility by @TejasGhatte in #4810
• fix: billing on failed responses stream requests anthropic and bedrock by @TejasGhatte in #4842
• feat(mcp): add per-MCP-server tool execution timeout by @Purvi09 in #4472
• feat: use schemas.SecretVar for virtual key values to support env store by @BearTS in #4817
• fix: honor model_parameters_url changes in config.json like pricing_url by @jeremym-tanium in #4864
• feat: add multi-value attribution cell with plural fallback for logs columns by @impoiler in #4865
• feat: add user_name, team_ids, team_names, customer_ids, customer_names, business_unit_ids, business_unit_names to log list select columns by @impoiler in #4866
• feat: latency info on errors by @TejasGhatte in #4867
• Added missing OpenAI responses methods for lifecycle related tasks by @17jmumford in #3125
• fix: small latency return fixes by @TejasGhatte in #4876
• feat: adds multiple teams / customers / bus to connectors by @roroghost17 in #4875
• fix: append datasheet models for incomplete list models call by @TejasGhatte in #4879
• refactor: simplify Responses lifecycle permissions to require explicit per-verb flags and expose them in UI by @Pratham-Mishra04 in #4880
• fix: skip O(N) reference refresh on request-time rate-limit/budget reset by @danpiths in #4883
• feat: add BedrockMantleKeyConfig support to key hashing, schema/table mapping, and sensitive field clearing by @Pratham-Mishra04 in #4886
• fix: cost for image generation or image edit streaming by @sammaji in #4802
• feat: adds deepseek provider by @sammaji in #4852
• chore: adds tests for deepseek provider by @sammaji in #4853
• chore: adds docs for deepseek provider by @sammaji in #4854
• fix: converts thinking to disabled if tool choice is required for deepseek by @sammaji in #4861
• feat: add expiry field to virtual keys by @Pratham-Mishra04 in #4887
• test: add Postman e2e collection and runner for virtual key expiry validation and enforcement by @Pratham-Mishra04 in #4888
• docs: add virtual key expiry support docs by @Pratham-Mishra04 in #4889
• ipv6 support by @akshaydeo in #4895
• docs: update benchmarking docs for new CLI flags, mocker/hitter tools, and revised result schema by @Pratham-Mishra04 in #4909
• fix: report cached_tokens as reads only per the OpenAI spec by @fus3r in #4906
• docs: clarify adaptive load balancing weight calculation and cross-node coordination by @Pratham-Mishra04 in #4744
• docs: add conflicting credentials troubleshooting and clarify MCP auth mode credential exclusivity by @Pratham-Mishra04 in #4910
• fix: surface Gemini batch inline responses from response field, not dest by @nnNyx in #4904
• fix: clear the stream close claim between attempts so streaming retries and fallbacks work by @fus3r in #4911
• [fix]: sanitize ErrorDetailsParsed so raw payloads honor disable_content_logging by @citrocat in #4873
• fix: decode URL-encoded governance team IDs in fetch/update/delete endpoints by @nnNyx in #4912
• fix: complete deferred LLM span on streaming goroutine exit by @TejasGhatte in #4885
• [fix]: sweep orphaned deferred spans in trace store TTL cleanup by @citrocat in #4869
• pr review skill by @akshaydeo in #4914
• fix: sanitize error details on log update path, set MCP raw-storage flag, remove redundant error serialization by @akshaydeo in #4913
• review skill updates by @akshaydeo in #4915
• fix: resolve semantic cache internal embedding keys like external requests by @nnNyx in #4903
• refactor: extract ClearContextForInternalRequest and extend it to clear body-transport state for internal embedding requests by @Pratham-Mishra04 in #4918
• fix: skip model check for responses lifecycle apis by @TejasGhatte in #4920
• fixes empty tool call result insertion failures by @akshaydeo in #4925
• fix(anthropic): consistent content_block indices for server tools on Claude Code passthrough streaming by @surki in #4890
• feat: extend Bedrock vendor-prefix pricing fallback to OpenAI, Google, and xAI models, and fold bedrock_mantle onto bedrock lookups by @Pratham-Mishra04 in #4924
• fix: adds support for external budget resolved for supplemental budgets not tracked against vk by @roroghost17 in #4825
• fix: stay on login page when SSO is configured but auth restart is pending by @BearTS in #4928
• fix: skip plaintext fallback lookup for secret-based virtual key values by @BearTS in #4927
• fix: tier costs evaluated via input tokens instead of total tokens by @TejasGhatte in #4917
• fix: stats for cancelled requests by @TejasGhatte in #4930
• fix: minor claude code fixes by @TejasGhatte in #4932
• fix: emit contentBlockStop events on the Bedrock ConverseStream egress by @fus3r in #4923
• clickhouse support for log_store by @akshaydeo in #4748
• hybdrid store with clickhouse by @akshaydeo in #4893
• chore: ui build fix by @TejasGhatte in #4933
• fix: set content type header consistently for responses api requests by @TejasGhatte in #4935
• feat: mark deprecated models instead of filtering them from API responses by @impoiler in #4936
• fix: idle timeout wiring in vertex path by @TejasGhatte in #4937
• fix: set SecretTypePlainText for plain-text JSON and non-prefixed secret values by @BearTS in #4946
• test: fix and update tests for tool_search round-trip, Anthropic passthrough flags, Gemini raw-body gating, and schema sync exclusions by @Pratham-Mishra04 in #4948
• test: additions to harness tests by @TejasGhatte in #4944
• fix: web fetch fixes by @TejasGhatte in #4945
• fix: gemini 2.5-pro thinking budget value by @TejasGhatte in #4947
• fix: guard setting tool call config in gemini by @TejasGhatte in #4959
• fix: round-trip anthropic redacted_thinking blocks on the chat completions surface by @fus3r in #4943
• fix: fixes error type setting in all integrations for Bedrock by @roroghost17 in #4958
• bug notice for v1.6.2 by @akshaydeo in #4961
• fix typo by @akshaydeo in #4962
• fix: propagate max_tokens from openai integration by @TejasGhatte in #4966
• chore: renames NewEnvVar to NewSecretVar in docs by @sammaji in #4955
• fix: adds file id for anthropic files integration and adds forwarding of content type
  by @sammaji in #4956
• chore: adds harness test cases for file apis in anthropic by @sammaji in #4967
• v1.6.3 changelogs by @akshaydeo in #4969
• fix: bedrock responses type error forwrding by @roroghost17 in #4968
• feat: add ClickHouse logs store, MCP OAuth2 server config, bedrock_mantle provider, virtual key expires_at, and per-client toolExecutionTimeout to Helm chart by @BearTS in #4960
• v1.6.3 by @akshaydeo in #4970
• fix: assert SecretTypePlainText value in redaction test instead of empty by @BearTS in #4971
• framework test case failure fix by @akshaydeo in #4972

New Contributors

• @mmacvicar made their first contribution in #4832
• @fus3r made their first contribution in #4906
• @citrocat made their first contribution in #4873
• @surki made their first contribution in #4890

Full Changelog: ent-v1.5.2-base...ent-v1.5.3-base