ltmoerdani · ltmoerdani · Aug 15, 2026 · Aug 13, 2026 · Aug 13, 2026 · Aug 14, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -4,8 +4,16 @@ All notable changes to the **OpenCode Go BYOK Provider** extension are documente
 
 ## [Unreleased]
 
+### Added
+
+- **`[Providers]` Configurable API base URL.** The extension no longer hardcodes the `opencode.ai` endpoints — it can now point at any compatible gateway via `opencodego.apiBaseUrl` (default `https://opencode.ai/zen/go/v1`) and `opencodezen.apiBaseUrl` (default `https://opencode.ai/zen/v1`). The extension derives the `/chat/completions`, `/messages`, `/responses`, `/models`, and (Go only) `/usage` routes from the configured base and applies them to every provider (normal chat, Agents window variants, inline completions, usage sync). Custom URLs are normalized and validated — non-`http(s)`, embedded credentials, query strings, and hashes are rejected and fall back to the default. Reload the window after changing the setting. Unit tests added for URL normalization and route construction.
+
+---
+
 ### Changed
 
+- **`[Providers]` Configurable API base URL.** The extension no longer hardcodes the `opencode.ai` endpoints — it can now point at any compatible gateway via `opencodego.apiBaseUrl` (default `https://opencode.ai/zen/go/v1`) and `opencodezen.apiBaseUrl` (default `https://opencode.ai/zen/v1`). The extension derives the `/chat/completions`, `/messages`, `/responses`, `/models`, and (Go only) `/usage` routes from the configured base and applies them to every provider (normal chat, Agents window variants, inline completions, usage sync). Custom URLs are normalized and validated — non-`http(s)`, embedded credentials, query strings, and hashes are rejected and fall back to the default. Reload the window after changing the setting. Unit tests added for URL normalization and route construction.
+
 - **`[Internal]` API keys are configured through the BYOK panel only.** The `OpenCode Go: Set API Key` / `OpenCode Zen: Set API Key` commands and the "Set / Clear API Key" menu items inside `Manage Provider` are removed — keys are entered once via **Chat: Manage Language Models → "+ Add Models"** (the native BYOK flow). `SecretStorage` is no longer a user-facing entry point; it stays as an internal per-vendor mirror (`opencodego.apiKey` / `opencodezen.apiKey`) that the BYOK resolution writes so agent-host variants and cold-start requests inherit the group key. Splitting the secret per vendor also fixes a latent collision where Go and Zen shared a single `opencodego.apiKey` and overwrote each other's key. `Refresh Models` / `Test Connection` now point at the BYOK flow when no key is configured instead of prompting for one.
 
 - **`[Internal]` Per-provider Thinking strategy classes + single config authority.** The thinking/reasoning system is refactored from one monolithic builder into a per-provider strategy (`src/thinking/`): an interface + factory (`provider.ts`), a shared base class, and one class per model family (`deepseek`, `glm`, `kimi`, `minimax`, `openai`, `qwen`, `mimo`, `fallback`). Each provider now owns its reasoning picker schema, its request-payload mapping, and whether its `reasoning_content` is surfaced as chat content. Configuration resolves from a **single authority** — the VS Code per-model configuration (model picker / Manage), with workspace settings and per-family defaults as fallbacks — instead of competing sources (workspace + modelConfiguration + a `globalState` shadow copy + defaults). The shadow copy is removed, so a thinking effort chosen for one model can no longer silently leak onto another model or override an explicit "Off". Model IDs are normalized to `effectiveModelId` (the `::sk-***` fp suffix is gone), which also stops the per-model settings group from being recreated on every pick. Request builders are split out of `extension.ts` into per-endpoint modules (`src/request/{types,schema,shared,openai,anthropic,google}.ts`). Windows tooling fixes: `scripts/lint.ts` runs npm `.cmd` shims through the shell and a new `.gitattributes` enforces LF normalization, so `npm run lint` (prettier + shellcheck) is green on Windows; `scripts/staged-lint.ts` and `isCwdInWorkspace` get the same treatment.

diff --git a/README.md b/README.md
@@ -381,30 +381,32 @@ Provider diagnostics also include the VS Code/extension versions, extension host
 
 ## 🔧 Settings
 
-| Setting                                   | Default  | Description                                                                                                            |
-| ----------------------------------------- | -------- | ---------------------------------------------------------------------------------------------------------------------- |
-| `opencodego.temperature`                  | `0.2`    | Sampling temperature (`0`–`2`)                                                                                         |
-| `opencodego.maxTokens`                    | `0`      | Max output token override (`0` = per-model max)                                                                        |
-| `opencodego.maxInputTokens`               | `0`      | Context window override (`0` = per-model default)                                                                      |
-| `opencodego.debugReasoning`               | `false`  | Log `reasoning_content` to Output panel                                                                                |
-| `opencodego.requestTimeoutSeconds`        | `600`    | Total request timeout                                                                                                  |
-| `opencodego.streamIdleTimeoutSeconds`     | `120`    | Cancel if stream goes idle                                                                                             |
-| `opencodego.showUsageStatusBar`           | `true`   | Show usage summary in status bar                                                                                       |
-| `opencodego.showProviderPrefix`           | `true`   | Include `OpenCode Go` / `OpenCode Zen` in model names                                                                  |
-| `opencodego.visionProxyWholeConversation` | `false`  | Vision proxy: describe the whole conversation instead of only the message with a new image (more context, more tokens) |
-| `opencodego.freeOnly`                     | `true`   | Zen: free models only. `false` = include paid                                                                          |
-| `opencodego.enabled`                      | `true`   | Register the OpenCode Go provider. `false` removes it from Language Models & every picker (keys kept)                  |
-| `opencodezen.enabled`                     | `true`   | Register the OpenCode Zen provider. `false` removes it from Language Models & every picker (keys kept)                 |
-| `opencodego.agentsWindow`                 | `true`   | Expose agent-host model variants (`targetChatSessionType`) for the Agents window                                       |
-| `opencodego.showAgentModelsInManagePanel` | `false`  | Show agent vendors in Manage Language Models panel                                                                     |
-| `opencodego.stripThinkTags`               | `"auto"` | Strip `<think>` tags (`never`/`auto`/`always`)                                                                         |
-| `opencodego.thinking.deepseek`            | `"off"`  | `off`/`low`/`medium`/`high`/`max`                                                                                      |
-| `opencodego.thinking.glm`                 | `"off"`  | `off`/`high`/`max`                                                                                                     |
-| `opencodego.thinking.kimi`                | `"off"`  | `on`/`off`                                                                                                             |
-| `opencodego.thinking.minimax`             | `"off"`  | `off`/`on`                                                                                                             |
-| `opencodego.thinking.mimo`                | `"off"`  | `off`/`low`/`medium`/`high`                                                                                            |
-| `opencodego.thinking.qwen`                | `"off"`  | `auto`/`on`/`off`                                                                                                      |
-| `opencodego.thinking.qwenBudget`          | `"auto"` | `auto`/`4096`/`16384`/`32768`/`81920`                                                                                  |
+| Setting                                   | Default                         | Description                                                                                                            |
+| ----------------------------------------- | ------------------------------- | ---------------------------------------------------------------------------------------------------------------------- |
+| `opencodego.apiBaseUrl`                   | `https://opencode.ai/zen/go/v1` | Base URL for a Go-compatible gateway; the extension appends the required API routes. Reload after changing.            |
+| `opencodezen.apiBaseUrl`                  | `https://opencode.ai/zen/v1`    | Base URL for a Zen-compatible gateway; the extension appends the required API routes. Reload after changing.           |
+| `opencodego.temperature`                  | `0.2`                           | Sampling temperature (`0`–`2`)                                                                                         |
+| `opencodego.maxTokens`                    | `0`                             | Max output token override (`0` = per-model max)                                                                        |
+| `opencodego.maxInputTokens`               | `0`                             | Context window override (`0` = per-model default)                                                                      |
+| `opencodego.debugReasoning`               | `false`                         | Log `reasoning_content` to Output panel                                                                                |
+| `opencodego.requestTimeoutSeconds`        | `600`                           | Total request timeout                                                                                                  |
+| `opencodego.streamIdleTimeoutSeconds`     | `120`                           | Cancel if stream goes idle                                                                                             |
+| `opencodego.showUsageStatusBar`           | `true`                          | Show usage summary in status bar                                                                                       |
+| `opencodego.showProviderPrefix`           | `true`                          | Include `OpenCode Go` / `OpenCode Zen` in model names                                                                  |
+| `opencodego.visionProxyWholeConversation` | `false`                         | Vision proxy: describe the whole conversation instead of only the message with a new image (more context, more tokens) |
+| `opencodego.freeOnly`                     | `true`                          | Zen: free models only. `false` = include paid                                                                          |
+| `opencodego.enabled`                      | `true`                          | Register the OpenCode Go provider. `false` removes it from Language Models & every picker (keys kept)                  |
+| `opencodezen.enabled`                     | `true`                          | Register the OpenCode Zen provider. `false` removes it from Language Models & every picker (keys kept)                 |
+| `opencodego.agentsWindow`                 | `true`                          | Expose agent-host model variants (`targetChatSessionType`) for the Agents window                                       |
+| `opencodego.showAgentModelsInManagePanel` | `false`                         | Show agent vendors in Manage Language Models panel                                                                     |
+| `opencodego.stripThinkTags`               | `"auto"`                        | Strip `<think>` tags (`never`/`auto`/`always`)                                                                         |
+| `opencodego.thinking.deepseek`            | `"off"`                         | `off`/`low`/`medium`/`high`/`max`                                                                                      |
+| `opencodego.thinking.glm`                 | `"off"`                         | `off`/`high`/`max`                                                                                                     |
+| `opencodego.thinking.kimi`                | `"off"`                         | `on`/`off`                                                                                                             |
+| `opencodego.thinking.minimax`             | `"off"`                         | `off`/`on`                                                                                                             |
+| `opencodego.thinking.mimo`                | `"off"`                         | `off`/`low`/`medium`/`high`                                                                                            |
+| `opencodego.thinking.qwen`                | `"off"`                         | `auto`/`on`/`off`                                                                                                      |
+| `opencodego.thinking.qwenBudget`          | `"auto"`                        | `auto`/`4096`/`16384`/`32768`/`81920`                                                                                  |
 
 <details>
 <summary><b>📜 Full settings reference with descriptions</b></summary>

diff --git a/package.json b/package.json
@@ -200,6 +200,16 @@
           "maximum": 2,
           "description": "Sampling temperature used for chat completions."
         },
+        "opencodego.apiBaseUrl": {
+          "type": "string",
+          "default": "https://opencode.ai/zen/go/v1",
+          "description": "Base URL for the OpenCode Go-compatible API. The extension appends /models, /chat/completions, /messages, /responses, and /usage. Requires a window reload after changing."
+        },
+        "opencodezen.apiBaseUrl": {
+          "type": "string",
+          "default": "https://opencode.ai/zen/v1",
+          "description": "Base URL for the OpenCode Zen-compatible API. The extension appends /models, /chat/completions, /messages, and /responses. Requires a window reload after changing."
+        },
         "opencodego.maxTokens": {
           "type": "number",
           "default": 0,

diff --git a/src/agentProvider.ts b/src/agentProvider.ts
@@ -0,0 +1,20 @@
+/** Build an agent-host variant while preserving the base provider endpoints. */
+export function providerVariant<T extends { vendor: string; displayName: string }, AgentVendor extends string, BaseVendor extends string>(
+  base: T,
+  agentVendor: AgentVendor,
+  displayName: string,
+  baseVendor: BaseVendor,
+): Omit<T, "vendor" | "displayName"> & {
+  vendor: AgentVendor;
+  displayName: string;
+  isAgentVariant: true;
+  baseVendor: BaseVendor;
+} {
+  return {
+    ...base,
+    vendor: agentVendor,
+    displayName,
+    isAgentVariant: true,
+    baseVendor,
+  };
+}
diff --git a/src/config.ts b/src/config.ts
@@ -45,6 +45,8 @@ export const SETTING_TEMPERATURE = "temperature";
 export const SETTING_MAX_TOKENS = "maxTokens";
 export const SETTING_MAX_INPUT_TOKENS = "maxInputTokens";
 export const SETTING_DEBUG_REASONING = "debugReasoning";
+/** Base URL setting key for the provider's OpenAI-compatible API. */
+export const SETTING_API_BASE_URL = "apiBaseUrl";
 export const SETTING_REQUEST_TIMEOUT_SECONDS = "requestTimeoutSeconds";
 export const SETTING_STREAM_IDLE_TIMEOUT_SECONDS = "streamIdleTimeoutSeconds";
 export const SETTING_STRIP_THINK_TAGS = "stripThinkTags";
@@ -115,6 +117,33 @@ export const MODEL_METADATA_CACHE_TTL_MS = 1 * 60 * 60 * 1000;
 export const DEFAULT_MODEL_CONTEXT_WINDOW = 262144;
 export const DEFAULT_MODEL_MAX_OUTPUT_TOKENS = 65536;
 
+// ─── Provider API endpoints ─────────────────────────────────────────────────
+
+/** Default OpenCode Go API base URL; can be overridden in VS Code settings. */
+export const DEFAULT_GO_API_BASE_URL = "https://opencode.ai/zen/go/v1";
+/** Default OpenCode Zen API base URL; can be overridden in VS Code settings. */
+export const DEFAULT_ZEN_API_BASE_URL = "https://opencode.ai/zen/v1";
+
+/** Normalize a configured API base URL, falling back when it is malformed. */
+export function normalizeApiBaseUrl(value: string, fallback: string): string {
+  const candidate = value.trim();
+  if (!candidate) return fallback;
+  try {
+    const url = new URL(candidate);
+    if ((url.protocol !== "http:" && url.protocol !== "https:") || url.username || url.password || url.search || url.hash) {
+      return fallback;
+    }
+    return url.toString().replace(/\/+$/, "");
+  } catch {
+    return fallback;
+  }
+}
+
+/** Append one API route to a normalized or user-supplied base URL. */
+export function appendApiPath(baseUrl: string, route: string): string {
+  return `${baseUrl.replace(/\/+$/, "")}/${route.replace(/^\/+/, "")}`;
+}
+
 // ─── Output budget / token-estimate margins ──────────────────────────────────
 
 /** Reserve for UI rendering so the advertised output never claims the full window. */

diff --git a/src/test/agentProvider.test.ts b/src/test/agentProvider.test.ts
@@ -0,0 +1,27 @@
+import assert from "node:assert/strict";
+import { describe, it } from "node:test";
+import { providerVariant } from "../agentProvider.js";
+
+describe("providerVariant", () => {
+  it("preserves configured endpoints and marks the provider as an agent variant", () => {
+    const base = {
+      vendor: "opencodego",
+      displayName: "OpenCode Go",
+      baseUrl: "https://example.test/custom/v1",
+      modelsUrl: "https://example.test/custom/v1/models",
+      chatCompletionsUrl: "https://example.test/custom/v1/chat/completions",
+      messagesUrl: "https://example.test/custom/v1/messages",
+      responsesUrl: "https://example.test/custom/v1/responses",
+    };
+
+    const agent = providerVariant(base, "opencodego-agent", "OpenCode Go (Agents)", "opencodego");
+
+    assert.equal(agent.isAgentVariant, true);
+    assert.equal(agent.baseVendor, "opencodego");
+    assert.equal(agent.vendor, "opencodego-agent");
+    assert.equal(agent.modelsUrl, base.modelsUrl);
+    assert.equal(agent.chatCompletionsUrl, base.chatCompletionsUrl);
+    assert.equal(agent.messagesUrl, base.messagesUrl);
+    assert.equal(agent.responsesUrl, base.responsesUrl);
+  });
+});