@proxysoul/soulforge 2.2.1 → 2.3.0

package/README.md

@@ -63,7 +63,7 @@ SoulForge doesn't work that way. On startup, it builds a **live dependency graph
 
 - **Lock-in mode.** Hides agent narration during work, shows only tool activity and the final answer. Toggle via `/lock-in` or config.
 - **Embedded Neovim.** Your actual config, plugins, and LSP servers. The AI works through the same editor you use. [Deep dive →](docs/architecture.md)
-- **17 providers.** Anthropic, OpenAI, Google, xAI, Groq, DeepSeek, Mistral, Amazon Bedrock, Fireworks, GitHub Copilot, GitHub Models, Ollama, OpenRouter, LLM Gateway, Vercel AI Gateway, Proxy, and any OpenAI-compatible API.
+- **18 providers.** Anthropic, OpenAI, Google, xAI, Groq, DeepSeek, Mistral, Amazon Bedrock, Fireworks, GitHub Copilot, GitHub Models, Ollama, LM Studio, OpenRouter, LLM Gateway, Vercel AI Gateway, Proxy, and any OpenAI-compatible API.
 - **Task router.** Assign different models to different jobs. Spark agents (explore/investigate) and ember agents (code edits) can each use different models. You pick what goes where. [Deep dive →](docs/architecture.md)
 - **Code execution (Smithy).** Sandboxed code execution via Anthropic's `code_execution` tool. The agent can run Python to process data, do calculations, or batch tool calls programmatically.
 - **User steering.** Type while the agent works. Messages queue up and reach the agent at the next step. [Deep dive →](docs/steering.md)
@@ -88,7 +88,7 @@ SoulForge doesn't work that way. On startup, it builds a **live dependency graph
 | **Task routing** | Per-task model assignment (spark, ember, web search, verify, desloppify, compact) | Single model | Single model | Per-agent model | Single model |
 | **Compound tools** | `read` (batch + surgical), `multi_edit` (atomic), `rename_symbol`, `move_symbol`, `rename_file`, `refactor`, `project` | Rename via LSP | — | — | — |
 | **Editor** | Embedded Neovim (your config, your plugins) | No editor | No editor | No editor | No editor |
-| **Providers** | 17 + custom OpenAI-compatible | Anthropic only | Multi-model | OpenAI only | 100+ LLMs |
+| **Providers** | 18 + custom OpenAI-compatible | Anthropic only | Multi-model | OpenAI only | 100+ LLMs |
 | **License** | BSL 1.1 (source-available) | Proprietary | Proprietary | Apache 2.0 | Apache 2.0 |
 
 > *Competitor features verified as of March 29, 2026. [Let us know](https://github.com/ProxySoul/soulforge/issues) if something's changed.*
@@ -215,6 +215,7 @@ soulforge --headless --diff "fix the bug"   # Show changed files
 | [**GitHub Copilot**](https://github.com/features/copilot) | OAuth token from IDE ([setup](docs/copilot-provider.md)) |
 | [**GitHub Models**](https://github.com/marketplace/models) | `GITHUB_MODELS_API_KEY` (PAT with `models:read`) |
 | [**Ollama**](https://ollama.ai) | Auto-detected |
+| [**LM Studio**](https://lmstudio.ai) | Auto-detected |
 | [**OpenRouter**](https://openrouter.ai) | `OPENROUTER_API_KEY` |
 | [**Vercel AI Gateway**](https://vercel.com/ai-gateway) | `AI_GATEWAY_API_KEY` |
 | [**Proxy**](https://github.com/router-for-me/CLIProxyAPI) | `PROXY_API_KEY` |
@@ -226,6 +227,10 @@ soulforge --headless --diff "fix the bug"   # Show changed files
 
 **GitHub Models**: Free playground API with per-token billing. Create a fine-grained PAT with `models:read` scope. Lower rate limits than Copilot.
 
+**Ollama**: Auto-detected at `localhost:11434`. Override with `OLLAMA_HOST=http://host:port`.
+
+**LM Studio**: Auto-detected at `localhost:1234`. Uses the [REST API v0](https://lmstudio.ai/docs/developer/rest/endpoints) for rich model data (context length, type filtering). Override with `LM_STUDIO_URL=http://host:port`. Optional auth: set `LM_API_TOKEN` if you've enabled authentication in LM Studio.
+
 Add custom providers in config, no code changes:
 
 ```json

package/dist/index.js

@@ -39260,7 +39260,7 @@ var package_default;
 var init_package = __esm(() => {
   package_default = {
     name: "@proxysoul/soulforge",
-    version: "2.2.1",
+    version: "2.3.0",
     description: "Graph-powered code intelligence \u2014 multi-agent coding with codebase-aware AI",
     repository: {
       type: "git",
@@ -39982,7 +39982,8 @@ function buildCustomProvider(config2) {
     id: config2.id,
     name: config2.name ?? config2.id,
     envVar,
-    icon: "\u25C7",
+    icon: "\uF29F",
+    asciiIcon: "\u25C7",
     custom: true,
     createModel(modelId) {
       const apiKey = envVar ? getProviderApiKey(envVar) ?? "" : "custom";
@@ -49575,6 +49576,75 @@ var init_llmgateway = __esm(() => {
   };
 });
 
+// src/core/llm/providers/lmstudio.ts
+function getBaseOrigin() {
+  return (process.env.LM_STUDIO_URL ?? "http://localhost:1234").replace(/\/+$/, "");
+}
+function openaiBase() {
+  return `${getBaseOrigin()}/v1`;
+}
+function restBase() {
+  return `${getBaseOrigin()}/api/v0`;
+}
+function getApiToken() {
+  return process.env.LM_API_TOKEN ?? "lm-studio";
+}
+function authHeaders() {
+  const token = getApiToken();
+  return token && token !== "lm-studio" ? {
+    Authorization: `Bearer ${token}`
+  } : {};
+}
+var lmstudio;
+var init_lmstudio = __esm(() => {
+  init_dist6();
+  lmstudio = {
+    id: "lmstudio",
+    name: "LM Studio",
+    envVar: "LM_API_TOKEN",
+    secretKey: "lm-api-token",
+    icon: "\uEA79",
+    asciiIcon: "L",
+    description: "Local models via LM Studio \u2014 no key needed",
+    createModel(modelId) {
+      const client = createOpenAI({
+        baseURL: openaiBase(),
+        apiKey: getApiToken()
+      });
+      return client.chat(modelId);
+    },
+    async fetchModels() {
+      const res = await fetch(`${restBase()}/models`, {
+        headers: authHeaders(),
+        signal: AbortSignal.timeout(3000)
+      });
+      if (!res.ok)
+        throw new Error(`LM Studio API ${String(res.status)}`);
+      const data = await res.json();
+      if (!Array.isArray(data.data))
+        return null;
+      return data.data.filter((m) => m.type === "llm" || m.type === "vlm").map((m) => ({
+        id: m.id,
+        name: m.id,
+        contextWindow: m.max_context_length
+      }));
+    },
+    fallbackModels: [],
+    async checkAvailability() {
+      try {
+        const res = await fetch(`${restBase()}/models`, {
+          headers: authHeaders(),
+          signal: AbortSignal.timeout(1000)
+        });
+        return res.ok;
+      } catch {
+        return false;
+      }
+    },
+    contextWindows: []
+  };
+});
+
 // node_modules/vercel-minimax-ai-provider/node_modules/@ai-sdk/anthropic/node_modules/@ai-sdk/provider/dist/index.mjs
 function getErrorMessage4(error48) {
   if (error48 == null) {
@@ -58676,6 +58746,9 @@ var init_mistral = __esm(() => {
 });
 
 // src/core/llm/providers/ollama.ts
+function getOllamaHost() {
+  return (process.env.OLLAMA_HOST ?? "http://localhost:11434").replace(/\/+$/, "");
+}
 var ollama;
 var init_ollama = __esm(() => {
   init_dist6();
@@ -58683,18 +58756,18 @@ var init_ollama = __esm(() => {
     id: "ollama",
     name: "Ollama",
     envVar: "",
-    icon: "\uD83E\uDD99",
-    asciiIcon: "\uD83E\uDD99",
+    icon: "\uEBA2",
+    asciiIcon: "O",
     description: "Local models \u2014 no key needed",
     createModel(modelId) {
       const client = createOpenAI({
-        baseURL: "http://localhost:11434/v1",
+        baseURL: `${getOllamaHost()}/v1`,
         apiKey: "ollama"
       });
       return client.chat(modelId);
     },
     async fetchModels() {
-      const res = await fetch("http://localhost:11434/api/tags");
+      const res = await fetch(`${getOllamaHost()}/api/tags`);
       if (!res.ok)
         throw new Error(`Ollama API ${String(res.status)}`);
       const data = await res.json();
@@ -58721,7 +58794,7 @@ var init_ollama = __esm(() => {
     }],
     async checkAvailability() {
       try {
-        const res = await fetch("http://localhost:11434/api/tags", {
+        const res = await fetch(`${getOllamaHost()}/api/tags`, {
           signal: AbortSignal.timeout(1000)
         });
         return res.ok;
@@ -77733,6 +77806,7 @@ __export(exports_providers, {
   ollama: () => ollama,
   mistral: () => mistral2,
   minimax: () => minimax2,
+  lmstudio: () => lmstudio,
   llmgateway: () => llmgateway2,
   groq: () => groq2,
   google: () => google2,
@@ -77793,6 +77867,7 @@ var init_providers = __esm(() => {
   init_google();
   init_groq();
   init_llmgateway();
+  init_lmstudio();
   init_minimax();
   init_mistral();
   init_ollama();
@@ -77811,6 +77886,7 @@ var init_providers = __esm(() => {
   init_google();
   init_groq();
   init_llmgateway();
+  init_lmstudio();
   init_minimax();
   init_mistral();
   init_ollama();
@@ -77819,7 +77895,7 @@ var init_providers = __esm(() => {
   init_proxy();
   init_vercel_gateway();
   init_xai();
-  BUILTIN_PROVIDERS = [llmgateway2, anthropic2, proxy2, vercelGatewayProvider, openai2, xai2, google2, groq2, deepseek2, mistral2, bedrock2, fireworks2, minimax2, copilot, githubModels, openrouter2, ollama];
+  BUILTIN_PROVIDERS = [llmgateway2, anthropic2, proxy2, vercelGatewayProvider, openai2, xai2, google2, groq2, deepseek2, mistral2, bedrock2, fireworks2, minimax2, copilot, githubModels, openrouter2, ollama, lmstudio];
   allProviders = [...BUILTIN_PROVIDERS];
   providerMap = new Map(allProviders.map((p) => [p.id, p]));
 });
@@ -448981,6 +449057,12 @@ function matchCopilotPricing(model) {
   }
   return;
 }
+function isModelLocal(modelId) {
+  const slash = modelId.indexOf("/");
+  if (slash < 0)
+    return false;
+  return LOCAL_PROVIDERS.has(modelId.slice(0, slash).toLowerCase());
+}
 function isModelFree(modelId) {
   const id = modelId.toLowerCase();
   if (id.endsWith(":free") || id.endsWith("-free"))
@@ -448995,7 +449077,7 @@ function isModelFree(modelId) {
 }
 function matchPricing(modelId) {
   const id = modelId.toLowerCase();
-  if (isModelFree(modelId))
+  if (isModelLocal(modelId) || isModelFree(modelId))
     return FREE_PRICING;
   if (id.startsWith("copilot/")) {
     const model = id.slice("copilot/".length);
@@ -449219,7 +449301,7 @@ function startMemoryPoll(intervalMs = 2000) {
     });
   }, intervalMs);
 }
-var MODEL_PRICING, FREE_PRICING, DEFAULT_PRICING, FREE, M025, M033, M1, M32, M30, COPILOT_PRICING, ZERO_USAGE, ZERO_PROCESS_RSS, useStatusBarStore, memPollStarted = false, memPollTimer = null;
+var MODEL_PRICING, FREE_PRICING, DEFAULT_PRICING, FREE, M025, M033, M1, M32, M30, COPILOT_PRICING, LOCAL_PROVIDERS, ZERO_USAGE, ZERO_PROCESS_RSS, useStatusBarStore, memPollStarted = false, memPollTimer = null;
 var init_statusbar = __esm(() => {
   init_esm();
   init_middleware();
@@ -449619,6 +449701,7 @@ var init_statusbar = __esm(() => {
     "claude-opus-4.6": M32,
     "claude-opus-4.6-fast": M30
   };
+  LOCAL_PROVIDERS = new Set(["ollama", "lmstudio"]);
   ZERO_USAGE = {
     prompt: 0,
     completion: 0,
@@ -475094,8 +475177,10 @@ function fmtCost(usd) {
     return `$${usd.toFixed(3)}`;
   return `$${usd.toFixed(2)}`;
 }
-function buildContent2(costCents, cacheHitPct, free) {
+function buildContent2(costCents, cacheHitPct, free, local) {
   const tk = getThemeTokens();
+  if (local)
+    return new StyledText([fg(tk.success)("Local")]);
   if (free)
     return new StyledText([fg(tk.success)("FREE")]);
   const cost = costCents / 100;
@@ -475110,11 +475195,13 @@ function TokenDisplay() {
   const cacheHitRef = import_react59.useRef(0);
   const currentCostRef = import_react59.useRef(0);
   const freeRef = import_react59.useRef(false);
+  const localRef = import_react59.useRef(false);
   import_react59.useEffect(() => useStatusBarStore.subscribe((state) => {
     const usage = state.tokenUsage;
     const breakdown = usage.modelBreakdown;
     const modelIds = Object.keys(breakdown);
-    freeRef.current = modelIds.length > 0 && modelIds.every((mid) => isModelFree(mid));
+    localRef.current = modelIds.length > 0 && modelIds.every((mid) => isModelLocal(mid));
+    freeRef.current = !localRef.current && modelIds.length > 0 && modelIds.every((mid_0) => isModelFree(mid_0));
     const rawCost = breakdown && modelIds.length > 0 ? computeTotalCostFromBreakdown(breakdown) : 0;
     costRef.current = Math.round(rawCost * 100);
     const totalInput = usage.prompt + usage.subagentInput + usage.cacheRead + usage.cacheWrite;
@@ -475128,7 +475215,7 @@ function TokenDisplay() {
       currentCostRef.current = approach2(currentCostRef.current, target);
       try {
         if (textRef.current)
-          textRef.current.content = buildContent2(currentCostRef.current, cacheHitRef.current, freeRef.current);
+          textRef.current.content = buildContent2(currentCostRef.current, cacheHitRef.current, freeRef.current, localRef.current);
       } catch {}
     }, STEP_MS2);
     return () => clearInterval(timer);
@@ -475136,7 +475223,7 @@ function TokenDisplay() {
   return /* @__PURE__ */ import_jsx_dev_runtime2.jsxDEV("text", {
     ref: textRef,
     truncate: true,
-    content: buildContent2(currentCostRef.current, cacheHitRef.current, freeRef.current)
+    content: buildContent2(currentCostRef.current, cacheHitRef.current, freeRef.current, localRef.current)
   }, undefined, false, undefined, this);
 }
 var import_react59, STEP_MS2 = 50, EASE2 = 0.35;
@@ -491963,9 +492050,10 @@ function StatusDashboard({
         innerW
       }, "t-total", false, undefined, this));
       const sortedBd = Object.entries(su.modelBreakdown ?? {}).sort(([midA, a2], [midB, b5]) => computeModelCost(midB, b5) - computeModelCost(midA, a2));
-      const allFree = sortedBd.length > 0 && sortedBd.every(([mid_0]) => isModelFree(mid_0));
+      const allLocal = sortedBd.length > 0 && sortedBd.every(([mid_0]) => isModelLocal(mid_0));
+      const allFree = !allLocal && sortedBd.length > 0 && sortedBd.every(([mid_1]) => isModelFree(mid_1));
       const totalCost = sortedBd.length > 0 ? computeTotalCostFromBreakdown(su.modelBreakdown ?? {}) : 0;
-      if (totalCost > 0 || allFree) {
+      if (totalCost > 0 || allFree || allLocal) {
         const fmtCost2 = (c) => c < 0.01 ? `${c.toFixed(3)}` : `${c.toFixed(2)}`;
         lines.push(/* @__PURE__ */ import_jsx_dev_runtime2.jsxDEV(Spacer, {
           innerW
@@ -491976,27 +492064,28 @@ function StatusDashboard({
           innerW
         }, "h-cost", false, undefined, this));
         const costLabelW = Math.min(30, innerW - 20);
-        for (const [mid_1, usage_0] of sortedBd) {
-          const free = isModelFree(mid_1);
-          const c_0 = computeModelCost(mid_1, usage_0);
-          if (c_0 <= 0 && !free)
+        for (const [mid_2, usage_0] of sortedBd) {
+          const local = isModelLocal(mid_2);
+          const free = !local && isModelFree(mid_2);
+          const c_0 = computeModelCost(mid_2, usage_0);
+          if (c_0 <= 0 && !free && !local)
             continue;
           const pct = totalCost > 0 ? Math.round(c_0 / totalCost * 100) : 0;
           const maxModelW = costLabelW - 4;
-          const shortId = mid_1.length > maxModelW ? `${mid_1.slice(0, maxModelW - 1)}\u2026` : mid_1;
+          const shortId = mid_2.length > maxModelW ? `${mid_2.slice(0, maxModelW - 1)}\u2026` : mid_2;
           lines.push(/* @__PURE__ */ import_jsx_dev_runtime2.jsxDEV(EntryRow, {
             label: `  ${shortId}`,
-            value: free ? "FREE" : `${fmtCost2(c_0)}  (${String(pct)}%)`,
-            valueColor: free ? t2.success : t2.textPrimary,
+            value: local ? "Local" : free ? "FREE" : `${fmtCost2(c_0)}  (${String(pct)}%)`,
+            valueColor: local || free ? t2.success : t2.textPrimary,
             labelW: costLabelW,
             rightAlign: true,
             innerW
-          }, `cost-${mid_1}`, false, undefined, this));
+          }, `cost-${mid_2}`, false, undefined, this));
         }
         lines.push(/* @__PURE__ */ import_jsx_dev_runtime2.jsxDEV(EntryRow, {
           label: "  Total",
-          value: allFree ? "FREE" : fmtCost2(totalCost),
-          valueColor: allFree ? t2.success : t2.warning,
+          value: allLocal ? "Local" : allFree ? "FREE" : fmtCost2(totalCost),
+          valueColor: allLocal || allFree ? t2.success : t2.warning,
           labelW: costLabelW,
           rightAlign: true,
           innerW
@@ -492012,7 +492101,9 @@ function StatusDashboard({
         innerW
       }, "h-tabs", false, undefined, this));
       const fmtCost_0 = (c_1, modelIds) => {
-        if (modelIds && modelIds.length > 0 && modelIds.every((mid_2) => isModelFree(mid_2)))
+        if (modelIds && modelIds.length > 0 && modelIds.every((mid_3) => isModelLocal(mid_3)))
+          return "Local";
+        if (modelIds && modelIds.length > 0 && modelIds.every((mid_4) => isModelFree(mid_4)))
           return "FREE";
         return c_1 <= 0 ? "\u2014" : c_1 < 0.01 ? `$${c_1.toFixed(3)}` : `$${c_1.toFixed(2)}`;
       };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@proxysoul/soulforge",
-  "version": "2.2.1",
+  "version": "2.3.0",
   "description": "Graph-powered code intelligence — multi-agent coding with codebase-aware AI",
   "repository": {
     "type": "git",