Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 17 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,23 @@
All notable changes to **billion-context** are documented here.
Versions follow the merge of a `*_release-v*` branch; CI publishes to npm on tag.

## [0.1.38] — 2026-08-13

### Fixes

- **OpenAI `reasoning_content` round-trip for thinking-mode models** (#129): OpenAI-compatible reasoning models (DeepSeek-R1, GLM-4.6 thinking, Qwen-QwQ) emit `reasoning_content` (chain-of-thought) and require it be echoed back on subsequent requests — without it the upstream returns HTTP 400 `The reasoning_content in the thinking mode must be passed back to the API.` bili's OpenAI adapter dropped it (Anthropic `thinking` and Responses paths already handled it). Now `openaiToCore`/`coreToOpenai` round-trip `reasoning_content` through a `contentType:"reasoning"` core message; the streaming adapter captures `delta.reasoning_content`; re-request reconstruction re-emits it so the model never sees a missing-CoT 400. Also fixed double-forwarding when a single chunk carried both `reasoning_content` and `content`.

## [0.1.37] — 2026-08-13

### Features

- **MITM domain auto-discovery** (#125): bili now reads client configs (`~/.zcode/v2/config.json`, `~/.codex/config.toml`, `~/.pi/agent/models.json`, `~/.claude/settings.json`) and auto-builds the MITM whitelist from all discovered HTTPS provider hostnames (mtime-cached, re-scanned on change). No more hardcoded domain assumptions — `open.bigmodel.cn`, `zcode.z.ai`, `api.z.ai` are now discovered, not baked in. Defaults reduced to the three binary-hardcoded endpoints (api.anthropic.com / api.openai.com / chatgpt.com) that have no config file to discover from.

### Fixes

- **Anthropic round-2 streaming framing (vertical-text bug)** (#126): after a proxy tool/compress re-request, round-2 streamed text rendered as vertical text (each ~2-char chunk on its own line). `runCompressLoop` round-2 text now carries the Anthropic `content_block_start/delta/stop` framing with the correct client index. Regression test added.
- **HTTP-proxy-mode absolute URLs** (#127): bili concatenated its default anthropic upstream with the full absolute request URL in HTTP-proxy mode (`https://api.anthropic.comhttp://127.0.0.1:18081/…`). Absolute URLs (e.g. a local model server) are now forwarded to the host in the URL instead of mangled.

## [0.1.36] — 2026-08-12

### Fixes
Expand Down
4 changes: 2 additions & 2 deletions package-lock.json

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

2 changes: 1 addition & 1 deletion package.json
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
{
"name": "billion-context",
"version": "0.1.37",
"version": "0.1.38",
"description": "Universal context-compression proxy for AI coding agents. Sits between any agent and its model API, rewriting Anthropic/OpenAI streams with acp-kernel compression. Any agent that can set a base URL (Claude Code, Codex, Cursor, Aider) works out of the box.",
"type": "module",
"main": "dist/index.js",
Expand Down
Loading