All pull requests

Pull requests

• fix(core): retry safe HTTP 429 responses in OpenAI passthrough

  #9315·n0namer opened Oct 29, 2026·

  ·
  1

• feat(ui/mcp): add registry for extra access grants in MCP server view sheet

  #9313·Pratham-Mishra04 opened Oct 29, 2026Collaborator·4 of 18 tasks·

  ·
  9

• feat(mcp): show per-caller discovery counts in the sessions table

  #9312·Pratham-Mishra04 opened Oct 29, 2026Collaborator·8 of 18 tasks·

  ·
  8

• feat(ui): split the MCP server sheet into a read-only view and a settings editor

  #9311·Pratham-Mishra04 opened Oct 29, 2026Collaborator·18 tasks·

  ·
  9

• feat(mcp): add per-user discovery test fixture MCP server example

  #9310·Pratham-Mishra04 opened Oct 29, 2026Collaborator·5 of 18 tasks·

  ·
  7

• perf(mcp): share one cached per-user discovery copy across callers with identical discovery

  #9309·Pratham-Mishra04 opened Oct 29, 2026Collaborator·4 of 18 tasks·

  ·
  8

• updates test scripts

  #9295·akshaydeo opened Oct 29, 2026Contributor·18 tasks·

  ·
  1

  2

• fix: routing rules for primary targets on 4xx

  #9292·akshaydeo opened Oct 29, 2026Contributor·8 of 18 tasks·

  ·
  1

  11

• fix: empty text document data loss in anthropic calls

  #9290·akshaydeo opened Oct 29, 2026Contributor·6 of 18 tasks·

  ·
  1

  6

• perf(openai): reduce request-body retention during SSE streaming

  #9287·G-XD opened Oct 29, 2026Contributor·8 of 18 tasks·

  ·
  4

• feat(redaction): redact streamed raw response text split across frames

  #9285·Madhuvod opened Oct 29, 2026Contributor·6 of 18 tasks·

  ·
  2

• fix(guardrails): attach raw stream events to their text delta and sync them across providers, replacing the stream text codec

  #9282·Madhuvod opened Oct 29, 2026Contributor·8 of 18 tasks·

  ·
  2

• docs: azure speech apis

  #9281·TejasGhatte opened Oct 29, 2026Collaborator·5 of 18 tasks·

  ·
  18

• docs: document the Kimchi provider and add it to the config schema, CI and harness

  #9280·arnasnn opened Oct 29, 2026·9 of 18 tasks·

  ·
  22

• feat: add Kimchi as a first-class provider

  #9277·arnasnn opened Oct 29, 2026·8 of 18 tasks·

  ·
  1

  7

• fix(gemini): forward chat completions seed to generationConfig.seed

  #9275·pcbeingused333 opened Oct 29, 2026Contributor·

  ·
  11

• feat: gateway-wide default for the request timeout (client.default_request_timeout)

  #9273·siaarzh opened Oct 29, 2026·8 of 18 tasks·

  ·
  1

• [fix]: Perplexity, Cohere and HuggingFace providers - do not remove mapped keys from the caller's ExtraParams

  #9267·pcbeingused333 opened Oct 29, 2026Contributor·3 of 11 tasks·

  ·
  6

• feat(vertex,gemini): stream Gemini 3+ tool-call arguments incrementally

  #9266·Disdjj opened Oct 29, 2026·7 of 18 tasks·

  ·
  1

  9

• fix(transports): return a JSON 404 for unmatched /api paths instead of the dashboard shell

  #9260·pcbeingused333 opened Oct 29, 2026Contributor·

  ·
  6

• [fix]: preserve Mistral non-streaming thinking content

  #9259·wozhenbang2004 opened Oct 29, 2026·8 of 10 tasks·

  ·
  1

  7

• [fix]: Anthropic provider - preserve requested web tool versions

  #9258·ykakade opened Oct 29, 2026·8 of 20 tasks·

  ·
  1

  1

• docs: provider level passthrough gating

  #9257·TejasGhatte opened Oct 28, 2026Collaborator·4 of 18 tasks·

  ·
  8

• Allow disabling stale-connection retries per request

  #9256·beckman9191 opened Oct 28, 2026·

  ·
  12

• fix(openai): read vLLM max_model_len as the model context length

  #9246·dougkirkley opened Oct 28, 2026·6 of 18 tasks·

  ·
  1

  2