refactor: remove fastgpt runtime dependency

Common/MAC_NODE_CONNECTIVITY_SPEC.md

@@ -2,17 +2,16 @@
 
 The Mac node should only do the following:
 
-1. Deploy FastGPT locally
-2. Ensure the cloud backend can reach FastGPT
+1. Deploy StoryForge collector-service locally
+2. Ensure the cloud backend can reach collector-service
 3. Maintain a private network connection to the server
-4. Provide the FastGPT endpoint to the backend
+4. Provide the collector-service endpoint to the backend
 
 Recommended ports:
 
-- FastGPT: 3000
-- MongoDB: 27017
-- PostgreSQL: 5432
-- Redis: 6379
-- MinIO: 9000
+- Collector Service: 8081
+- n8n: 5670
+- Local Model API: 8317
+- ASR: 8088
 
-FastGPT must not be exposed to the public internet directly.
+The local admin/control surfaces must not be exposed to the public internet directly.

Common/STORYFORGE_MAC_AI_NODE_TASKS.md

@@ -2,13 +2,11 @@ You are responsible for the StoryForge Mac AI node.
 
 Tasks:
 
-- Deploy FastGPT using Docker.
+- Deploy StoryForge runtime services on the Mac node.
 - Services:
-  - FastGPT
-  - MongoDB
-  - PostgreSQL + pgvector
-  - Redis
-  - MinIO
+  - collector-service
+  - n8n
+  - cli-proxy-api
 - Build collector-service in Python.
 - Collector features:
   - yt-dlp video download