@elvatis_com/openclaw-cli-bridge-elvatis 2.8.1 → 2.8.3

package/README.md

@@ -2,7 +2,7 @@
 
 > OpenClaw plugin that bridges locally installed AI CLIs (Codex, Gemini, Claude Code, OpenCode, Pi) as model providers — with slash commands for instant model switching, restore, health testing, and model listing.
 
-**Current version:** `2.8.1`
+**Current version:** `2.8.3`
 
 ---
 
@@ -406,6 +406,13 @@ npm run ci          # lint + typecheck + test
 
 ## Changelog
 
+### v2.8.3
+- **fix:** transparent fallback — responses now always report the originally requested model, preventing clients from "sticking" on Haiku after a single Sonnet timeout. Fallback still happens internally for resilience, but is invisible to the client.
+
+### v2.8.2
+- **fix:** increase Sonnet-4-6 and Opus-4-6 base timeout 300s→420s (7 min) to prevent premature timeout→Haiku fallback on large webchat sessions
+- **fix:** increase Haiku-4-5 base timeout 90s→120s for better reliability as fallback model
+
 ### v2.8.1
 - **fix:** increase Sonnet-4-6 base timeout from 180s to 300s to prevent premature SIGTERM kills causing FailoverError fallback to gpt-5.2-codex
 

package/SKILL.md

@@ -68,4 +68,4 @@ On gateway restart, if any session has expired, a **WhatsApp alert** is sent aut
 
 See `README.md` for full configuration reference and architecture diagram.
 
-**Version:** 2.8.1
+**Version:** 2.8.3

package/openclaw.plugin.json

@@ -2,7 +2,7 @@
   "id": "openclaw-cli-bridge-elvatis",
   "slug": "openclaw-cli-bridge-elvatis",
   "name": "OpenClaw CLI Bridge",
-  "version": "2.8.1",
+  "version": "2.8.3",
   "license": "MIT",
   "description": "Phase 1: openai-codex auth bridge. Phase 2: local HTTP proxy routing model calls through gemini/claude CLIs (vllm provider).",
   "providers": [

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@elvatis_com/openclaw-cli-bridge-elvatis",
-  "version": "2.8.1",
+  "version": "2.8.3",
   "description": "Bridges gemini, claude, and codex CLI tools as OpenClaw model providers. Reads existing CLI auth without re-login.",
   "type": "module",
   "openclaw": {

package/src/config.ts

@@ -86,14 +86,14 @@ export const PROVIDER_SESSION_SWEEP_MS = 10 * 60 * 1_000; // 10 min
  * Override via `modelTimeouts` in plugin config.
  *
  * Strategy:
- *   - Heavy/agentic models (Opus, GPT-5.4): 5 min — need time for tool use
- *   - Standard interactive (Sonnet, Pro, GPT-5.3): 3 min
- *   - Fast/lightweight (Haiku, Flash, Mini): 90s
+ *   - Heavy/agentic models (Opus, GPT-5.4): 7 min — need time for tool use + large sessions
+ *   - Standard interactive (Sonnet, Pro, GPT-5.3): 7 min — prevents premature fallback to Haiku
+ *   - Fast/lightweight (Haiku, Flash, Mini): 120s
  */
 export const DEFAULT_MODEL_TIMEOUTS: Record<string, number> = {
-  "cli-claude/claude-opus-4-6":        300_000,  // 5 min
-  "cli-claude/claude-sonnet-4-6":      300_000,  // 5 min — match idleTimeoutSeconds
-  "cli-claude/claude-haiku-4-5":        90_000,  // 90s
+  "cli-claude/claude-opus-4-6":        420_000,  // 7 min
+  "cli-claude/claude-sonnet-4-6":      420_000,  // 7 min — prevent timeout→Haiku fallback on large sessions
+  "cli-claude/claude-haiku-4-5":       120_000,  // 2 min
   "cli-gemini/gemini-2.5-pro":         300_000,  // 5 min — image generation needs more time
   "cli-gemini/gemini-2.5-flash":       180_000,  // 3 min
   "cli-gemini/gemini-3-pro-preview":   300_000,  // 5 min — image generation needs more time

package/src/proxy-server.ts

@@ -104,9 +104,9 @@ export interface ProxyServerOptions {
    *
    * Example:
    *   {
-   *     "cli-claude/claude-sonnet-4-6": 180_000,   // 3 min for interactive chat
-   *     "cli-claude/claude-opus-4-6":   300_000,    // 5 min for heavy tasks
-   *     "cli-claude/claude-haiku-4-5":  90_000,     // 90s for fast responses
+   *     "cli-claude/claude-sonnet-4-6": 420_000,   // 7 min for interactive chat
+   *     "cli-claude/claude-opus-4-6":   420_000,    // 7 min for heavy tasks
+   *     "cli-claude/claude-haiku-4-5":  120_000,    // 2 min for fast responses
    *   }
    *
    * When not set for a model, falls back to proxyTimeoutMs (default 300s base).
@@ -807,7 +807,7 @@ async function handleRequest(
           const fbCompTokens = estimateTokens(result.content ?? "");
           metrics.recordRequest(fallbackModel, Date.now() - fallbackStart, true, estPromptTokens, fbCompTokens);
           usedModel = fallbackModel;
-          opts.log(`[cli-bridge] fallback to ${fallbackModel} succeeded`);
+          opts.log(`[cli-bridge] fallback to ${fallbackModel} succeeded (response will report original model: ${model})`);
         } catch (fallbackErr) {
           metrics.recordRequest(fallbackModel, Date.now() - fallbackStart, false, estPromptTokens);
           const fallbackMsg = (fallbackErr as Error).message;
@@ -851,7 +851,7 @@ async function handleRequest(
         const toolCalls = result.tool_calls!;
         // Role chunk with all tool_calls (name + empty arguments)
         sendSseChunk(res, {
-          id, created, model: usedModel,
+          id, created, model,
           delta: {
             role: "assistant",
             tool_calls: toolCalls.map((tc, idx) => ({
@@ -864,7 +864,7 @@ async function handleRequest(
         // Arguments chunks (one per tool call)
         for (let idx = 0; idx < toolCalls.length; idx++) {
           sendSseChunk(res, {
-            id, created, model: usedModel,
+            id, created, model,
             delta: {
               tool_calls: [{ index: idx, function: { arguments: toolCalls[idx].function.arguments } }],
             },
@@ -872,20 +872,20 @@ async function handleRequest(
           });
         }
         // Stop chunk
-        sendSseChunk(res, { id, created, model: usedModel, delta: {}, finish_reason: "tool_calls" });
+        sendSseChunk(res, { id, created, model, delta: {}, finish_reason: "tool_calls" });
       } else {
         // Standard text streaming
-        sendSseChunk(res, { id, created, model: usedModel, delta: { role: "assistant" }, finish_reason: null });
+        sendSseChunk(res, { id, created, model, delta: { role: "assistant" }, finish_reason: null });
         const content = result.content ?? "";
         const chunkSize = 50;
         for (let i = 0; i < content.length; i += chunkSize) {
           sendSseChunk(res, {
-            id, created, model: usedModel,
+            id, created, model,
             delta: { content: content.slice(i, i + chunkSize) },
             finish_reason: null,
           });
         }
-        sendSseChunk(res, { id, created, model: usedModel, delta: {}, finish_reason: "stop" });
+        sendSseChunk(res, { id, created, model, delta: {}, finish_reason: "stop" });
       }
 
       res.write("data: [DONE]\n\n");
@@ -903,7 +903,7 @@ async function handleRequest(
         id,
         object: "chat.completion",
         created,
-        model: usedModel,
+        model,
         choices: [
           {
             index: 0,

package/test/config.test.ts

@@ -61,9 +61,9 @@ describe("config.ts exports", () => {
   });
 
   it("exports per-model timeouts for all major models", () => {
-    expect(DEFAULT_MODEL_TIMEOUTS["cli-claude/claude-opus-4-6"]).toBe(300_000);
-    expect(DEFAULT_MODEL_TIMEOUTS["cli-claude/claude-sonnet-4-6"]).toBe(300_000);
-    expect(DEFAULT_MODEL_TIMEOUTS["cli-claude/claude-haiku-4-5"]).toBe(90_000);
+    expect(DEFAULT_MODEL_TIMEOUTS["cli-claude/claude-opus-4-6"]).toBe(420_000);
+    expect(DEFAULT_MODEL_TIMEOUTS["cli-claude/claude-sonnet-4-6"]).toBe(420_000);
+    expect(DEFAULT_MODEL_TIMEOUTS["cli-claude/claude-haiku-4-5"]).toBe(120_000);
     expect(DEFAULT_MODEL_TIMEOUTS["cli-gemini/gemini-2.5-pro"]).toBe(300_000);
     expect(DEFAULT_MODEL_TIMEOUTS["cli-gemini/gemini-2.5-flash"]).toBe(180_000);
     expect(DEFAULT_MODEL_TIMEOUTS["openai-codex/gpt-5.4"]).toBe(300_000);