ur-agent 1.82.0 → 1.82.2

package/docs/CONFIGURATION.md

@@ -74,7 +74,7 @@ UR-Nexus supports official provider access paths only:
 
 - Explicit API providers: OpenAI, Anthropic, Gemini, OpenRouter, and
   OpenAI-compatible endpoints.
-- Local/server providers: Ollama, LM Studio, llama.cpp, and vLLM OpenAI-compatible
+- Local/server providers: Ollama, LM Studio, llama.cpp, vLLM, and Unsloth OpenAI-compatible
   server mode.
 - Subscription CLI providers: Codex CLI, Claude Code CLI, Gemini CLI, and
   Antigravity where officially supported. These dispatch turns through the
@@ -117,6 +117,7 @@ ur config set provider anthropic-api
 ur config set provider gemini-api
 ur config set provider openrouter
 ur config set provider openai-compatible
+ur config set provider unsloth
 ur provider doctor agy
 ur config set provider.fallback ollama
 ur config set model <model>
@@ -130,7 +131,7 @@ recovery command; changing providers remains an explicit user action.
 
 Provider values accept canonical IDs and common aliases. Examples:
 `openai-api`, `anthropic-api`, `gemini-api`, `openrouter`, `ollama`,
-`lmstudio`, `LM Studio`, `llama.cpp`, `vllm`, and the subscription CLI
+`lmstudio`, `LM Studio`, `llama.cpp`, `vllm`, `unsloth` (`Unsloth Studio`), and the subscription CLI
 providers `codex-cli` (`chatgpt`), `claude-code-cli` (`claude`), `gemini-cli`
 (`gemini`), and `antigravity-cli` (`agy`). Values with spaces should be quoted
 in shell commands.
@@ -143,6 +144,11 @@ incompatible saved model instead of silently carrying it across providers. The
 saved provider/model pair controls the runtime backend for the next agent
 request; Ollama is only used when `ollama` is the selected provider.
 
+The configured `base_url` is provider-scoped. Setting an address while vLLM is
+active does not replace the saved Ollama, llama.cpp, or Unsloth address;
+returning to any provider restores its own URL. Legacy `provider.baseUrl`
+settings are migrated to the old active provider when the first switch occurs.
+
 In the model step, Up/Down browses, Left/Right changes the focused model's
 supported effort level, Enter confirms, Ctrl+R refreshes the catalog, and Esc
 returns to providers. OpenRouter entries show pricing tier, context size,
@@ -150,6 +156,9 @@ tool/reasoning capability, compact names, and the exact ID for the focused
 entry. Its catalog is fetched fresh whenever opened; a failed refresh never
 silently displays cached entries. API-provider secret
 entry stays on one masked row and stores the value through the keychain flow.
+Ollama and llama.cpp capabilities are loaded lazily for the focused model from
+`/api/show` and `/props`, respectively, so the arrow selector reflects the
+actual model rather than a provider-wide guess.
 
 The same provider-first picker is mandatory on the first interactive run in a
 workspace with no model in `.ur/settings.json` or `.ur/settings.local.json`.
@@ -179,8 +188,16 @@ OPENAI_COMPATIBLE_API_KEY=...
 ANTHROPIC_API_KEY=...
 GEMINI_API_KEY=...
 OPENROUTER_API_KEY=...
+UNSLOTH_API_KEY=...
 ```
 
+Unsloth is an inference-provider integration only. Start Unsloth Studio and
+load the model outside UR, connect its generated key with `ur connect unsloth`,
+then select a model discovered from `http://localhost:8888/v1` (or your
+configured `base_url`). UR always sends `enable_tools: false`; standard model
+function calls continue through UR's own permission, sandbox, and verifier
+flow, while Unsloth's server-side tools remain disabled.
+
 ### OpenAI Responses transport
 
 OpenAI uses Chat Completions unless the Responses transport is selected

package/docs/TROUBLESHOOTING.md

@@ -198,10 +198,29 @@ ur provider doctor <provider-id>
 ```sh
 curl http://localhost:11434/api/tags        # Ollama
 curl http://localhost:1234/v1/models        # LM Studio (llama.cpp: 8080, vLLM: 8000)
+curl -H "Authorization: Bearer $UNSLOTH_API_KEY" http://localhost:8888/v1/models
 ur connect openai-api                       # store an API key securely
 ur provider doctor
 ```
 
+### Unsloth is selected but unavailable
+
+- Likely cause: Studio is not running, no model is loaded, its generated API
+  key is not connected, or `base_url` does not end at the compatible API root.
+- Fix: start/load Unsloth outside UR, connect its key, and inspect discovery.
+
+```sh
+echo "$UNSLOTH_API_KEY" | ur connect unsloth
+ur config set provider unsloth
+ur config set base_url http://localhost:8888/v1
+ur provider doctor unsloth
+ur provider models unsloth --json
+```
+
+UR does not start or manage Unsloth. It also disables Unsloth server-side tools
+on every request; tool execution shown by UR is handled by UR's own guarded
+tool loop.
+
 ### Local server unreachable
 
 - Likely cause: wrong `base_url` or the server is not running.
@@ -212,6 +231,10 @@ ur config set base_url http://localhost:11434
 ur provider doctor
 ```
 
+Addresses are saved per provider. If the doctor probes an unexpected URL,
+select that provider first and run `ur config get base_url`; changing vLLM's
+address no longer overwrites Ollama, llama.cpp, or Unsloth.
+
 ## Sessions and workflows
 
 ### No visible progress in scripts

package/docs/USAGE.md

@@ -162,6 +162,7 @@ ur config set provider anthropic-api
 ur config set provider gemini-api
 ur config set provider openrouter
 ur config set provider openai-compatible
+ur config set provider unsloth
 ur config set model <model>
 ur config set base_url <url>
 ur config set provider.fallback ollama
@@ -206,17 +207,25 @@ kind, external CLI usage, native tool/streaming support, and the boundary text.
 
 Provider values accept canonical IDs and common aliases. For example,
 `openai-api`, `anthropic-api`, `gemini-api`, `openrouter`, `ollama`,
-`lmstudio`, `llama.cpp`, and `vllm` are UR-native runtime providers, and
+`lmstudio`, `llama.cpp`, `vllm`, and `unsloth` are UR-native runtime providers, and
 `codex-cli` (`chatgpt`), `claude-code-cli` (`claude`), `gemini-cli` (`gemini`),
 and `antigravity-cli` (`agy`) are subscription CLI providers.
 
 API modes are explicit. Keys are read from a key stored via
 `ur connect <provider>` (OS keychain) or from the environment variables
 `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, `GEMINI_API_KEY`, and
-`OPENROUTER_API_KEY`. Subscription CLIs are optional, never required
+`OPENROUTER_API_KEY`, and `UNSLOTH_API_KEY`. Subscription CLIs are optional, never required
 dependencies, and never used as a silent fallback. UR-Nexus never scrapes
 browser sessions, extracts OAuth tokens, or bypasses provider restrictions.
 OpenAI-compatible local or cloud endpoints use `base_url` plus `model`.
+Unsloth defaults to `http://localhost:8888/v1`, requires its Studio API key,
+and is inference-only: UR does not manage Unsloth and disables its server-side
+tools while retaining standard function calls inside UR's guarded tool loop.
+
+UR stores `base_url` per provider. You can set different addresses for
+Ollama, llama.cpp, vLLM, and Unsloth once, then switch providers without
+re-entering any of them. `ur config get base_url` always reports the address
+for the currently active provider.
 
 Use `/model` in an interactive session to select provider first and model
 second. OpenAI API, Claude API, Gemini API, OpenRouter, Ollama, and

package/docs/VALIDATION.md

@@ -19,7 +19,7 @@ You need:
 
 ```sh
 ur --version
-# expected for this release: "1.82.0 (UR-Nexus)"
+# expected for this release: "1.82.2 (UR-Nexus)"
 ```
 
 ### 0.0 Redteam mode and Reverse Skills (1.81.0)

package/docs/providers.md

@@ -41,6 +41,7 @@ multimodal input, external CLI boundary, and sandbox scope:
 | LM Studio | local/server | UR-native | no | yes | yes | yes | UR Bash/File sandbox | `openai-compatible:lmstudio` | local OpenAI-compatible server |
 | llama.cpp | local/server | UR-native | no | yes | yes | yes | UR Bash/File sandbox | `openai-compatible:llama.cpp` | local OpenAI-compatible server |
 | vLLM | local/server | UR-native | no | yes | yes | yes | UR Bash/File sandbox | `openai-compatible:vllm` | OpenAI-compatible server |
+| Unsloth | local/server | UR-native | no | yes | yes | model-dependent | UR Bash/File sandbox | `openai-compatible:unsloth` | authenticated user-run Unsloth Studio endpoint (`UNSLOTH_API_KEY`) |
 | Codex CLI | subscription | subscription-cli | yes | no | no | no† | UR-run tools/output only† | `subscription-cli:codex` | official Codex CLI login |
 | Claude Code | subscription | subscription-cli | yes | no | no | no† | UR-run tools/output only† | `subscription-cli:claude-code` | official Claude Code CLI login |
 | Gemini CLI | subscription | subscription-cli | yes | no | no | no† | UR-run tools/output only† | `subscription-cli:gemini` | official Gemini Code Assist login |
@@ -122,6 +123,7 @@ ur config set provider anthropic-api
 ur config set provider gemini-api
 ur config set provider openrouter
 ur config set provider openai-compatible
+ur config set provider unsloth
 ur config set model <model>
 ur provider select-model <provider> <model> --json
 ur config set base_url <url>
@@ -136,6 +138,11 @@ The fallback setting is a recovery hint for `ur provider doctor`; it does not
 route a failed request to another provider. Review the failure and use
 `ur config set provider <id>` to switch explicitly.
 
+`base_url` is stored for the provider that is active when the command runs.
+Each provider retains its own address across `/provider`, `/model`, and CLI
+switches. The legacy single `provider.baseUrl` field remains readable and is
+migrated to the previously active provider on the first switch.
+
 OpenAI API uses Chat Completions by default. `openai_transport responses` is an
 explicit opt-in to the native Responses adapter; it defaults to `store=false`
 and supports semantic streaming, background polling/cancellation, WebSocket
@@ -165,6 +172,17 @@ indicator, active-work spinner, SDK settings response, and provider request all
 use the same resolved value. If a provider advertises only boolean thinking,
 UR does not invent a graded effort selector.
 
+For Ollama, UR lazily reads the focused model's `/api/show` capabilities and
+sends the selected level through native `think`. Kimi K3 uses
+`low|high|max`; GPT-OSS uses `low|medium|high`; other models advertising
+`thinking` use Ollama's current `low|medium|high|max` contract. Direct OpenAI,
+Anthropic, and Gemini models use curated model-specific ladders from their
+official documentation; live discovery rows are merged with those contracts.
+See [Ollama thinking](https://docs.ollama.com/capabilities/thinking),
+[OpenAI model guidance](https://developers.openai.com/api/docs/guides/latest-model),
+[Claude effort](https://platform.claude.com/docs/en/build-with-claude/effort),
+and [Gemini thinking](https://ai.google.dev/gemini-api/docs/thinking).
+
 The provider-first `/model` picker supports the same control directly: use
 Left/Right to move through the effort levels advertised by the focused model,
 then Enter to apply the model and effort together. OpenRouter's live catalog
@@ -309,7 +327,7 @@ ur config set provider anthropic-api
 | --- | --- | --- |
 | API providers (openai-api, anthropic-api, gemini-api) | Live discovery from the provider's `/models` endpoint using your connected key (curated fallback until connected) | live |
 | OpenRouter | Fresh `/models` discovery every time its picker opens; no cached-list fallback in the picker | live |
-| Local/server providers (ollama, lmstudio, llama.cpp, vllm) | Dynamic discovery from the selected provider endpoint | live |
+| Local/server providers (ollama, lmstudio, llama.cpp, vllm, unsloth) | Dynamic discovery from the selected provider endpoint | live |
 | OpenAI-compatible | Dynamic discovery from configured endpoint | live |
 | Subscription CLIs (codex-cli, claude-code-cli, gemini-cli, antigravity-cli) | Curated list (the official CLIs expose no models API); first-class in `/model`, dispatched via the official CLI. External CLI behavior depends on the vendor CLI. Log in with `ur auth <provider>` | static |
 
@@ -332,6 +350,7 @@ ur config set provider anthropic-api
 - `lmstudio` — LM Studio OpenAI-compatible server
 - `llama.cpp` — llama.cpp server mode
 - `vllm` — vLLM server
+- `unsloth` — authenticated Unsloth Studio server; UR uses it for inference only
 
 **Important:**
 - A ChatGPT/Claude/Gemini subscription does NOT give API access
@@ -462,6 +481,7 @@ Provider config and doctor commands accept canonical IDs and common aliases:
 | `lmstudio` | `LM Studio`, `lm-studio` |
 | `llama.cpp` | `llama cpp`, `llamacpp`, `llama-cpp` |
 | `vllm` | `vllm server` |
+| `unsloth` | `Unsloth Studio`, `unsloth server`, `unsloth local` |
 
 `ur provider doctor` checks the selected provider. It reports installed/missing
 CLIs, official login status where available, API key presence for API providers,
@@ -496,6 +516,7 @@ OPENAI_COMPATIBLE_API_KEY=...
 ANTHROPIC_API_KEY=...
 GEMINI_API_KEY=...
 OPENROUTER_API_KEY=...
+UNSLOTH_API_KEY=...
 ```
 
 OpenAI-compatible endpoints can point at local or cloud endpoints:
@@ -516,6 +537,32 @@ Local/server providers use their normal endpoints:
 - LM Studio: `http://localhost:1234/v1`
 - llama.cpp server mode: `http://localhost:8080/v1`
 - vLLM server mode: `http://localhost:8000/v1`
+- Unsloth Studio: `http://localhost:8888/v1`
+
+### Unsloth provider-only mode
+
+UR connects to a user-run Unsloth Studio inference server through its official
+OpenAI-compatible API. It does not import the Unsloth Python package, launch or
+update Studio, train or convert models, manage GPUs, or load a model. Start and
+load Unsloth separately, then connect the generated Studio key:
+
+```sh
+ur config set provider unsloth
+echo "$UNSLOTH_API_KEY" | ur connect unsloth
+ur provider doctor unsloth
+ur provider models unsloth --json
+# choose one of the discovered model IDs with /model
+```
+
+The default endpoint is `http://localhost:8888/v1`; override it with
+`ur config set base_url <url>`. Authentication is mandatory. Every Unsloth
+request sets `enable_tools: false`, including streaming requests. The model may
+still return standard OpenAI function calls, but execution remains exclusively
+inside UR's provenance gate, permissions, sandbox, and verifier. This prevents
+Unsloth Studio's optional server-side web/code tools from becoming a second,
+uncontrolled agent runtime. See the official
+[Unsloth Studio announcement](https://github.com/unslothai/unsloth/discussions/5285)
+and [Unsloth repository](https://github.com/unslothai/unsloth).
 
 Ollama allows up to 15 minutes for response headers so a cold model load or
 large prefill can begin. After headers, local and `:cloud` models use a

package/documentation/index.html

@@ -45,7 +45,7 @@
     <main id="content" class="content">
       <header class="topbar">
         <div>
-          <p class="eyebrow">Version 1.82.0</p>
+          <p class="eyebrow">Version 1.82.2</p>
           <h1>UR-Nexus Documentation</h1>
           <p class="lead">A practical, tutorial-style reference for installing, configuring, automating, extending, and operating UR-Nexus.</p>
         </div>

package/extensions/jetbrains-ur/build.gradle.kts

@@ -7,7 +7,7 @@ plugins {
 }
 
 group = "dev.urnexus"
-version = "1.82.0"
+version = "1.82.2"
 
 repositories {
     mavenCentral()

package/extensions/vscode-ur-inline-diffs/package.json

@@ -2,7 +2,7 @@
   "name": "ur-inline-diffs",
   "displayName": "UR Inline Diffs",
   "description": "Review, apply, and reject UR inline diff bundles from .ur/ide/diffs inside VS Code.",
-  "version": "1.82.0",
+  "version": "1.82.2",
   "publisher": "ur-nexus",
   "engines": {
     "vscode": "^1.92.0"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ur-agent",
-  "version": "1.82.0",
+  "version": "1.82.2",
   "description": "UR-Nexus — autonomous engineering workflow engine (plan, execute, test, verify, document, benchmark, reproduce)",
   "type": "module",
   "packageManager": "bun@1.3.14",