Merge pull request #160 from esokullu/main

15.2

Author: esokullu

CHANGELOG.md

@@ -6,6 +6,19 @@ This changelog was generated from the repository Git history and release tags. V
 
 ## [Unreleased]
 
+## [15.2.0] - 2026-06-22
+
+### Added
+- Jan, vLLM, and SGLang as built-in local providers (Chrome + Firefox). All three use OpenAI-compatible `/v1` endpoints (Jan on port 1337, vLLM on port 8000, SGLang on port 30000), support model listing via `/v1/models`, accept an optional API key for auth-enabled servers, and default to enabled with vision on and a 16 K context window.
+
+### Changed
+- Onboarding local-model detection copy now lists Jan, vLLM, and SGLang alongside LM Studio, Ollama, and llama.cpp.
+- LLM request-timeout settings description and provider info panel updated to cover all six local backends.
+- Updated documentation (README, architecture docs, providers guide) to reflect the expanded local-provider lineup.
+
+### Tests
+- Added coverage for `categoryFor` and `listProviderModels` with Jan, vLLM, and SGLang — including auth header forwarding and model-list deduplication — and for `_defaultConfigs` asserting all three new providers are present, enabled, local-categorized, and localhost-defaulted.
+
 ## [15.1.1] - 2026-06-22
 
 ### Changed

README.md

@@ -18,7 +18,7 @@ Open-source AI browser agent for Chrome and Firefox. Chat with any web page, aut
 - **Continue from Limit** — When the agent hits the step limit, click Continue to keep going
 - **Multi-Provider LLM** — Supports local and cloud models:
   - **WebBrain Cloud 1.0** (cloud, default) — Built-in managed cloud option; no local setup required
-  - **llama.cpp** (local) — No API key needed. Also **Ollama** and **LM Studio**
+  - **llama.cpp** (local) — No API key needed. Also **Ollama**, **LM Studio**, **Jan**, **vLLM**, and **SGLang**
   - **OpenAI** (GPT-5.5, etc.)
   - **Anthropic Claude** (native API)
   - **Google Gemini**, **Mistral AI**, **DeepSeek**, **xAI Grok**, **Groq**
@@ -68,6 +68,13 @@ llama-server -m your-model.gguf --port 8080
 # Or using Ollama (OpenAI-compatible)
 ollama serve
 # Then set base URL to http://localhost:11434/v1 in settings
+
+# Or using Jan (OpenAI-compatible)
+# Start Jan's local API server and use http://localhost:1337/v1
+
+# Or using vLLM / SGLang (OpenAI-compatible)
+vllm serve your-model --port 8000
+python -m sglang.launch_server --model-path your-model --port 30000
 ```
 
 > **Context window:** For reliable agent runs, load a local model with **at least a 16k-token context window** (the usable minimum). 8k can work with **Compact mode** enabled (Settings → per-provider checkbox); 4k is too small to hold the system prompt + tool schemas. WebBrain auto-compacts the conversation as it nears the window — it assumes 16k for local models unless you set an explicit context size, so give the model server (e.g. `llama-server -c 16384`) enough room.
@@ -97,6 +104,9 @@ Click the gear icon or go to the extension's Options page to configure:
 | llama.cpp | `http://localhost:8080` | Not needed | (your loaded model) |
 | Ollama | `http://localhost:11434/v1` | Not needed | (your loaded model) |
 | LM Studio | `http://localhost:1234/v1` | Not needed | (your loaded model) |
+| Jan | `http://localhost:1337/v1` | Not needed | (your loaded model) |
+| vLLM | `http://localhost:8000/v1` | Optional | (your served model) |
+| SGLang | `http://localhost:30000/v1` | Optional | (your served model) |
 | OpenAI | `https://api.openai.com/v1` | Required | gpt-5.5 |
 | Anthropic Claude | `https://api.anthropic.com` | Required | claude-sonnet-4-6 |
 | Google Gemini | `https://generativelanguage.googleapis.com/v1beta/openai` | Required | gemini-3.1-flash |
@@ -180,7 +190,7 @@ Deeper docs live in [`docs/`](docs/): [architecture](docs/architecture.md), [sit
 | `solve_captcha` | -- | Yes | Yes | Solve CAPTCHAs via CapSolver API (optional, requires API key) |
 | `done` | Yes | Yes | Yes | Signal task completion |
 
-**Compact mode** is a reduced tool set + shorter system prompt designed for small local models (2B-8B). In both Chrome and Firefox builds, it cuts the Act-mode schema from 40+ tools to about 20, reducing decision surface and hallucination. Enable it per-provider in Settings (checkbox on llama.cpp, Ollama, LM Studio; off by default).
+**Compact mode** is a reduced tool set + shorter system prompt designed for small local models (2B-8B). In both Chrome and Firefox builds, it cuts the Act-mode schema from 40+ tools to about 20, reducing decision surface and hallucination. Enable it per-provider in Settings (checkbox on local providers; off by default).
 
 > **Shadow DOM note:** The accessibility tree only traverses light DOM. On Web Component-heavy pages (Stripe, Salesforce, Shopify), use `get_interactive_elements` (pierces open shadow roots) or `get_shadow_dom` / `shadow_dom_query` for targeted reads.
 

dist/webbrain-chrome-15.0.0.zip dist/webbrain-chrome-15.2.0.zipdist/webbrain-chrome-15.0.0.zip renamed to dist/webbrain-chrome-15.2.0.zip

@@ -13,7 +13,7 @@ So the question this document answers is: **what is the agent equivalent of the
 ## 2. System overview & trust boundaries
 
 - **Extension (Manifest V3).** The agent loop, prompt assembly, and tool dispatch run in the extension's standard MV3 sandbox.
-- **Local model process.** llama.cpp (via LM Studio / Ollama) runs as a *separate* process and is reached over `localhost` HTTP. No custom binaries, no elevated privileges; the model itself has only the extension's permissions, indirectly.
+- **Local model process.** llama.cpp, Ollama, LM Studio, Jan, vLLM, or SGLang runs as a *separate* process and is reached over `localhost` HTTP. No custom binaries, no elevated privileges; the model itself has only the extension's permissions, indirectly.
 - **Automation surface.** Page reads and actions are performed through the extension APIs and, for richer control, CDP/debugger automation.
 - **Cloud option.** The same agent can target a cloud model instead of the local one.
 

dist/webbrain-firefox-15.0.0.zip dist/webbrain-firefox-15.2.0.zipdist/webbrain-firefox-15.0.0.zip renamed to dist/webbrain-firefox-15.2.0.zip

@@ -25,7 +25,7 @@ The user's message, the current page content (AX tree, screenshot, or extracted
 The user chooses their provider in Settings. Options include:
 
 - **Cloud providers**: OpenAI, Anthropic, Google Gemini, Mistral, DeepSeek, xAI, Groq, OpenRouter, etc. — data leaves the user's machine for these
-- **Local providers**: llama.cpp, Ollama, LM Studio — data stays on the user's machine
+- **Local providers**: llama.cpp, Ollama, LM Studio, Jan, vLLM, SGLang — data stays on the user's machine
 
 The extension itself never receives or stores user data on any remote server.
 

docs/THREAT-MODEL.md

@@ -38,6 +38,9 @@ class BaseLLMProvider {
 | `llamacpp` | `llamacpp` | local | (loaded model) | Yes (default on) |
 | `ollama` | `openai` | local | (loaded model) | Yes (default on) |
 | `lmstudio` | `openai` | local | (loaded model) | Yes (default on) |
+| `jan` | `openai` | local | (loaded model) | Yes (default on) |
+| `vllm` | `openai` | local | (loaded model) | Yes (default on) |
+| `sglang` | `openai` | local | (loaded model) | Yes (default on) |
 | `openai` | `openai` | cloud | `gpt-5.5` | Model-name regex |
 | `anthropic` | `anthropic` | cloud | `claude-sonnet-4-6` | Model-name regex |
 | `claude_subscription` | `anthropic_oauth` | cloud | `claude-sonnet-4-6` | Yes |
@@ -53,13 +56,17 @@ class BaseLLMProvider {
 
 ### Local Providers
 
-Three local providers are enabled by default with no API key needed:
+Six local providers are enabled by default with no API key needed unless the
+local server was started with auth:
 
 - **llama.cpp**: `http://localhost:8080` — runs `llama-server -m model.gguf`
 - **Ollama**: `http://localhost:11434/v1` — `ollama serve`
 - **LM Studio**: `http://localhost:1234/v1` — LM Studio's local inference server
+- **Jan**: `http://localhost:1337/v1` — Jan's local OpenAI-compatible API server
+- **vLLM**: `http://localhost:8000/v1` — vLLM's OpenAI-compatible server
+- **SGLang**: `http://localhost:30000/v1` — SGLang's OpenAI-compatible server
 
-All three default `supportsVision: true` since most models loaded locally in 2026 are multimodal.
+All six default `supportsVision: true` since most models loaded locally in 2026 are multimodal.
 
 **Context window.** Load local models with **at least a 16k-token context window** for reliable agent runs — that's the usable minimum. 8k can work with Compact mode enabled; 4k is too small to hold the system prompt + tool schemas. The agent reads the window from `provider.contextWindow` (`providers/base.js`) to drive auto-compaction; when a provider config doesn't set `contextWindow`, local providers default to a conservative **16k** (cloud/router default to 128k). Set `config.contextWindow` explicitly to match a larger local window, and make sure the model server is actually started with that much context (e.g. `llama-server -c 16384`).
 
@@ -74,7 +81,7 @@ filters the exposed tools through `COMPACT_TOOL_NAMES`; Ask mode is unchanged.
 | OpenAI-compatible | Regex against model name (`gpt-4o`, `gpt-5`, `claude-3`, `claude-sonnet-4`, `gemini-2.0-flash`, etc.) |
 | Anthropic | `claude-(3\|sonnet-4\|opus-4)` patterns |
 | llama.cpp | Explicit `supportsVision` config toggle |
-| Ollama / LM Studio | Explicit `supportsVision` config toggle (via OpenAI provider) |
+| Ollama / LM Studio / Jan / vLLM / SGLang | Explicit `supportsVision` config toggle (via OpenAI provider) |
 
 ### Anthropic Conversion
 

docs/privacy-and-data-flow.md

@@ -1,7 +1,7 @@
 {
   "manifest_version": 3,
   "name": "WebBrain",
-  "version": "15.1.1",
+  "version": "15.2.0",
   "description": "Open-source AI browser agent — chat with pages, automate tasks, multi-provider LLM support.",
   "permissions": [
     "sidePanel",

docs/providers-and-models.md

@@ -1,6 +1,6 @@
 {
   "name": "webbrain",
-  "version": "15.1.1",
+  "version": "15.2.0",
   "description": "Open-source AI browser agent — chat with pages, automate tasks, multi-provider LLM support.",
   "private": true,
   "type": "module",