clarity-ai 6.5.3 → 6.5.5

package/CHANGELOG.md

@@ -2,6 +2,22 @@
 
 ---
 
+## 6.5.5 (2026-06-06)
+
+### Fix model routing — repo prefix stripped by regex
+- Fixed `MODEL_ROUTES` keys to match `Clarity-flash-weights` / `Clarity-heavy-weights` after `callAI` strips the provider AND repo prefixes
+- `npm install -g clarity-ai@latest` now works out of the box
+
+## 6.5.4 (2026-06-06)
+
+### Docker SDK + Local GGUF Inference
+- All 6 Spaces converted to `sdk: docker` with FastAPI + `llama-cpp-python`
+- Models run locally on each Space (no HF Inference API dependency)
+- Sync FastAPI handlers for compatibility; lazy model load on first request
+- Pre-built CPU wheels for `llama-cpp-python` via `--extra-index-url`
+- Spaces made public — no auth header needed
+- Default model: Qwen2.5-1.5B-Instruct-Q4_K_M (Flash & Heavy)
+
 ## 6.5.2 (2026-06-06)
 
 ### CoT + MoE Space Architecture

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "clarity-ai",
-  "version": "6.5.3",
-  "description": "CLARITY — terminal AI agent with Clarity Flash 14B",
+  "version": "6.5.5",
+  "description": "CLARITY — terminal AI agent with local GGUF inference on HF Spaces",
   "type": "module",
   "bin": {
     "clarity": "bin/clarity.js"

package/src/config/models.js

@@ -1,6 +1,6 @@
 export const ALL_MODELS = [
-  { id: 'huggingface/Universal-618/Clarity-flash-weights', provider: 'huggingface', label: 'Clarity Flash 14B', badge: '128K' },
-  { id: 'huggingface/Universal-618/Clarity-heavy-weights', provider: 'huggingface', label: 'Clarity Heavy 20B', badge: 'MoE' },
+  { id: 'huggingface/Universal-618/Clarity-flash-weights', provider: 'huggingface', label: 'Clarity Flash (Qwen 2.5 1.5B)', badge: 'Free' },
+  { id: 'huggingface/Universal-618/Clarity-heavy-weights', provider: 'huggingface', label: 'Clarity Heavy (Qwen 2.5 1.5B)', badge: 'Free' },
   { id: 'groq/llama-3.3-70b-versatile', provider: 'groq', label: 'Llama 3.3 70B Versatile', badge: null },
   { id: 'groq/llama-3.1-8b-instant', provider: 'groq', label: 'Llama 3.1 8B Instant', badge: 'Fast' },
   { id: 'groq/llama-4-scout-17b-16e-instruct', provider: 'groq', label: 'Llama 4 Scout 17B', badge: null },

package/src/providers/index.js

@@ -2,13 +2,13 @@ import { getKey } from '../config/keys.js';
 import { streamResponse } from './streaming.js';
 
 const MODEL_ROUTES = {
-  'Universal-618/Clarity-flash-weights': [
-    process.env.HUGGINGFACE_ENDPOINT || 'https://universal-618-clarity-2.hf.space/v1/chat/completions',
-    'https://universal-618-clarity-4.hf.space/v1/chat/completions',
-    'https://universal-618-clarity-5.hf.space/v1/chat/completions',
+  'Clarity-flash-weights': [
+    process.env.CLARITY_INFERENCE_URL || 'https://universal-618-clarity-main.hf.space/v1/chat/completions',
+    'https://universal-618-clarity-2.hf.space/v1/chat/completions',
+    'https://universal-618-clarity-3.hf.space/v1/chat/completions',
   ],
-  'Universal-618/Clarity-heavy-weights': [
-    process.env.HUGGINGFACE_ENDPOINT || 'https://universal-618-clarity-3.hf.space/v1/chat/completions',
+  'Clarity-heavy-weights': [
+    process.env.CLARITY_INFERENCE_URL || 'https://universal-618-clarity-4.hf.space/v1/chat/completions',
     'https://universal-618-clarity-5.hf.space/v1/chat/completions',
     'https://universal-618-clarity-6.hf.space/v1/chat/completions',
   ],

package/src/providers/streaming.js

@@ -26,7 +26,7 @@ export async function* streamResponse(endpoint, body, apiKey, extraHeaders = {},
       method: 'POST',
       headers: {
         'Content-Type': 'application/json',
-        'Authorization': 'Bearer ' + apiKey,
+        ...(apiKey && apiKey !== 'none' ? { 'Authorization': 'Bearer ' + apiKey } : {}),
         ...extraHeaders,
       },
       body: JSON.stringify({ ...body, stream: true }),