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 v1.5.2
​
Bifrost(HTTP)
1.5.2

​
✨ Features

  • Push/Pull Telemetry Toggling - Push-based and pull-based telemetry can now be toggled separately, and plugin hot-reload is fixed (#3433)
  • MCP Tool Manager Config Hash - mcp.tool_manager_config is now included in the client config hash and synced on reload (#3432)

​
🐞 Fixed

  • OTEL Metric Fixes - Resolved issues in the OTEL plugin metrics path (#3439)
  • Gemini Raw Request Scoping - Raw-request handling is now applied only for the Gemini provider (#3437)
  • Gemini Fallback Propagation - Fixed Gemini fallback propagation in the GenAI integration (#3338) (thanks @Javtor!)
  • Anthropic Advisor Model Passthrough - Fixed model passthrough prefix stripping in the advisor tool for Anthropic (#3420)
  • Empty Text/Signature Messages on Bedrock - Drop messages with empty text or signature and convert thinking blocks for OpenAI Bedrock models (#3221)
  • Anthropic Trailing Assistant Messages - Drop the last assistant message for Anthropic models and convert unsupported reasoning effort values for Mistral (#3203)
  • Compat Defaults Enabled - All compat plugin settings are now enabled by default (#3202)
  • System-Only Message Role Conversion - Convert role system to role user when only a system message is present for non-OpenAI models (#3200)
  • Compat CachePoint Drop - Compat plugin now drops cachePoint for unsupported Bedrock models and non-Bedrock models (#3154)
​
Core
1.5.10
  • fix: include mcp.tool_manager_config in client config hash (#3432)
  • fix: gemini fallback propagation in GenAI integration (#3338) (thanks @Javtor!)
  • fix: drops messages with empty text or signature and adds conversion of thinking block for openai bedrock models (#3221)
  • fix: drops last assistant messages for anthropic models and converts unsupported reasoning effort values for mistral (#3203)
  • fix: converts role system to role user if only system message is present for non-openai models (#3200)
  • fix: fixes model passthrough prefix stripping in advisor tool for anthropic (#3420)
​
Framework
1.3.10
  • fix: sync mcp.tool_manager_config on client config reload (#3432)
  • fix: enables all compat plugin settings by default (#3202)
  • feat: mistral reasoning effort entries in modelcatalog (#3203)
  • chore: compat plugin cachePoint drop catalog entries (#3154)
​
compat
0.1.9
  • fix: drops messages with empty text or signature and adds conversion of thinking block for openai bedrock models (#3221)
  • fix: drops last assistant messages for anthropic models and converts unsupported reasoning effort values for mistral (#3203)
  • fix: enables all compat plugin settings by default (#3202)
  • fix: converts role system to role user if only system message is present for non-openai models (#3200)
  • fix: drops cachePoint for unsupported bedrock models and non-bedrock models (#3154)
​
governance
1.5.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
jsonparser
1.5.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
logging
1.5.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
maxim
1.6.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
mocker
1.5.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
otel
1.2.10
  • fix: otel metric fixes (#3439)
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
prompts
1.0.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
semanticcache
1.5.10
  • chore: upgraded core to v1.5.10 and framework to v1.3.10
​
telemetry
1.5.10
  • feat: supports toggling push-based and pull-based separately and fixes hot-reload of plugin (#3433)
  • fix: otel metric fixes (#3439)
  • chore: upgraded core to v1.5.10 and framework to v1.3.10