Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.getbifrost.ai/llms.txt

Use this file to discover all available pages before exploring further.

npx -y @maximhq/bifrost --transport-version v1.5.1
Bifrost(HTTP)
1.5.1

✨ Features

  • Azure Entra Auth - Added client_id, client_secret, tenant_id, scopes fields to the Azure transport schema and made api_version optional (#3284)
  • UBI9 Docker Builds - Added UBI9 amd64/arm64 Docker build and multi-arch manifest jobs to the release pipeline (#3324)
  • Configurable Matview Refresh - Added matviewRefreshInterval config option and bumped the bifrost chart to 2.1.15 (#3329, #3386)
  • Per-dimension Matviews - Split mv_logs_filterdata into per-dimension matviews and added filterdata response caching with single-flight (#3288)
  • Dimension-scoped Filters - Added ?dimensions= parameter for filter data endpoints with a time-window hybrid matview gate and scoped dimension fetching for log and MCP filter sidebars (#3327, #3328)
  • x-bifrost- Response Headers* - Surface routed identity (provider, model, key alias, etc.) as x-bifrost-* response headers (#3280) (thanks @Edward-Upton!)
  • Normalized Model Names - ListModels response now includes a normalized_name parameter (#3372)
  • Azure Blob Batch Storage - Added Azure blob storage fields for the batch API (#3318)
  • WS Tickets Without Storage - WebSocket tickets are now issued without backing storage (#3361)
  • MCP Log Detail & Object Storage - Added MCP log detail endpoint with object-storage payload offload, batch logging for MCP, and detail view in the UI (#3350, #3351, #3352, #3353, #3354)
  • Retry Count Metric - Added a number-of-retries metric to OTEL and telemetry plugins (#3392)
  • Cached Token Details in OTEL - Added cached token details for the Responses API in OTEL exports (#3293)
  • Bedrock Region from Model Name - Region is now extracted from the model name when passing through Bedrock models (#3301)
  • ListModels Removed From Rate Limit/Quota - Governance no longer invokes ListModels for rate limit and quota checks (#3304)
  • JSON Log Detail Rendering - JSON text content is now rendered as formatted code in the log detail view (#3373)
  • x-operation-id Approval - Added approval flow for operation IDs (#3360)
  • Prometheus Plugin Cleanup - Removed system metrics from the prometheus plugin and added additional types (#3306)

🐞 Fixed

  • PostgreSQL \u0000 Sanitization - Sanitize \u0000 from JSON before PostgreSQL jsonb storage and list queries (#3332)
  • Lock-safe Matview Maintenance - Use batched updates, CONCURRENTLY indexes, and advisory locks for matview and migration maintenance (#3380)
  • pg_advisory_lock Retry+Timeout - Replaced blocking pg_advisory_lock with retry+timeout to prevent indefinite startup hangs (#3308)
  • Empty MCP Config - Return an empty MCP config instead of nil when no clients are found (#3387)
  • Gemini Image Size Casing - Removed lowercase normalization on image size in Gemini image requests (#3385)
  • Bedrock Reasoning Display - Reasoning display shorthand and set display=summarized for Responses on Bedrock (#3379)
  • Gemini JSON Schema for Vertex - Normalize JSON Schema union types for Vertex compatibility (#3150)
  • RDB Deadlock Fixes - Resolved deadlocks in the rate-limit DB path (#3376)
  • xhigh Effort Mapping - Added xhigh effort mapping for GPT 5.4+ (#3369)
  • Plugin Trace-Attribute Flow - Updated context.SetTraceAttribute flow across all plugins (#3368)
  • Opus 4.7 Reasoning Default - Default display=summarized for Opus 4.7 reasoning (#3375)
  • Gemini Video Download - Fixed video download for the Gemini provider (#3366)
  • Claude Code Native Converters Fallback - Fallback to native converters for Claude Code when an output format is set on the Vertex messages API (#3321)
  • Gemini tokenCount Always Included - Always include tokenCount in Gemini modality usage details (#3349) (thanks @R-droid101!)
  • Tool Choice Any in GenAI - Handle tool_choice=any in the GenAI roundtrip (#3315)
  • OTEL anyOf Schema - OTEL plugin now uses anyOf instead of oneOf to correctly validate host:port format for the gRPC schema (#3297)
  • scim_config Keycloak Alignment - Aligned scim_config schema with the documented Keycloak provider (#3346) (thanks @lornest!)
  • Docker Volume Path - Aligned Docker volume with the configured app directory (#3348) (thanks @EzgiTastan!)
  • OAuth Flow Preservation - Don’t clear oauthFlow on OAuth error (#3326)
  • MCP OAuth Window Null - Fixed MCP OAuth window null issue (#3359)
  • Governance Video/Passthrough Types - Updated video and passthrough request types in governance that do not require a model field (#3283)
  • Custom Provider Sheet Layout - Custom provider sheet header and footer are now sticky with proper padding and a scrollable content area (#3285)
  • Empty Filter Section Visibility - Prevent filter sections from hiding when opened with empty data (#3258)
  • Config JSON Client Credentials - Restored client id and secret in the config JSON output (#3284)
Core
1.5.9
  • fix: remove lowercase on image size in gemini image requests (#3385)
  • fix: reasoning display short hand and set display summarized in responses bedrock (#3379)
  • fix(gemini): normalize JSON Schema union types for Vertex compatibility (#3150)
  • feat: xhigh effort mapping for GPT 5.4+ (#3369)
  • chore: updates context.SetTraceAttribute flow for all plugins (#3368)
  • fix: default to display summarized in opus 4.7 reasoning (#3375)
  • feat: adds normalized_name parameter to the response of ListModels (#3372)
  • feat: fields for azure blob storage in batch api (#3318)
  • fix: video download for gemini provider (#3366)
  • feat: extracts region from model name while passing bedrock models (#3301)
  • feat: removes ListModels call from rate limit and quota check (#3304)
  • fix: always include tokenCount in Gemini modality usage details (#3349) (thanks @R-droid101!)
  • fix: handle tool choice any in genai roundtrip (#3315)
  • feat: adds cached token details in Responses API in OTEL exports (#3293)
  • chore: harness test related changes (#3391)
Framework
1.3.9
  • fix: sanitize \u0000 from JSON before PostgreSQL jsonb storage and list queries (#3332)
  • feat: add configurable matview_refresh_interval for PostgreSQL logs store (#3329)
  • feat: add custom filter values and scoped dimension fetching for logs and MCP filter sidebars (#3328)
  • fix: use batched updates, CONCURRENTLY indexes, and advisory locks for lock-safe matview and migration maintenance (#3380)
  • feat: add ?dimensions= param to filter data endpoints and use time-window hybrid matview gate for aggregate counts (#3327)
  • perf: split mv_logs_filterdata into per-dimension matviews and add filterdata response caching with single-flight (#3288)
  • fix: return empty MCP config instead of nil when no clients found (#3387)
  • fix: rdb deadlock fixes (#3376)
  • chore: updates context.SetTraceAttribute flow for all plugins (#3368)
  • feat: ws-tickets without storage (#3361)
  • feat: add batch logging to mcp logs similar to llm logs (#3354)
  • feat: offload MCP logs to object storage (#3353)
  • feat: add MCP log object payload helpers (#3352)
  • feat: adds cached token details in Responses API in OTEL exports (#3293)
  • fix: replace blocking pg_advisory_lock with retry+timeout to prevent indefinite startup hangs (#3308)
compat
0.1.8
  • chore: upgraded core to v1.5.9 and framework to v1.3.9
governance
1.5.9
  • fix: rdb deadlock fixes (#3376)
  • feat: removes ListModels call from rate limit and quota check (#3304)
  • fix: update video and passthrough request types in governance which do not require model field (#3283)
jsonparser
1.5.9
  • chore: upgraded core to v1.5.9 and framework to v1.3.9
logging
1.5.9
  • feat: add batch logging to mcp logs similar to llm logs (#3354)
  • feat: add MCP log detail endpoint (#3350)
maxim
1.6.9
  • chore: upgraded core to v1.5.9 and framework to v1.3.9
mocker
1.5.9
  • chore: upgraded core to v1.5.9 and framework to v1.3.9
otel
1.2.9
  • feat: add number of retries as a metric (#3392)
  • feat: adds cached token details in Responses API in OTEL exports (#3293)
prompts
1.0.9
  • chore: upgraded core to v1.5.9 and framework to v1.3.9
semanticcache
1.5.9
  • chore: upgraded core to v1.5.9 and framework to v1.3.9
telemetry
1.5.9
  • feat: add number of retries as a metric (#3392)
  • feat: removes system metrics from prometheus plugin and adds some more needed types (#3306)