src/albedo/openai_api · albedo · Tangled

small gleam coding and persistent agent daemon with a detachable cli and a canvas ui forked from okami.mom/albedo

albedo src albedo openai_api

16 files

nekomimi.pet · mistral: add a provider for Mistral's own API with key accounts and Vibe browser sign-in · 2w ago

albedo_openai_json.erl

openai_api: send reasoning_content on every assistant message once the history carries it

albedo_openai_pool.erl

openai_api: fast paths for chat tool fragments and claude deltas, requests wait for a draining connection, parallel and tool-call benchmark

albedo_openai_transport.erl

openai_api: read SSE event ids and retry, and open GET requests

albedo_sse_bytes.erl

openai_api: native utf-8 check and single-byte newline search in SSE framing (-39% reduce time), offline stream benchmark

chat: accept openrouter's role-bearing finish trailer; add openrouter to /login

refactor(gleam): clean up code and enforce focused lint checks

gemini: field-reader fast paths for vertex and cloud code assist text chunks, gemini mock routes

reasoning.gleam

chat: accept openrouter's role-bearing finish trailer; add openrouter to /login

refactor: remove unused internal code

mistral: add a provider for Mistral's own API with key accounts and Vibe browser sign-in

responses.gleam

responses: read an error event's message at the top or nested under error

openai_api: read SSE event ids and retry, and open GET requests

stream_http.gleam

bedrock: share one streaming HTTP pump between SSE and Converse

Refactor Gleam and Erlang code across daemon, harness, and providers

transport.gleam

openai_api: read SSE event ids and retry, and open GET requests

mistral: add a provider for Mistral's own API with key accounts and Vibe browser sign-in