czero-cc · nicoloval · Aug 16, 2025 · Aug 15, 2025 · Aug 15, 2025 · Aug 15, 2025
diff --git a/QUICKSTART.md b/QUICKSTART.md
@@ -14,6 +14,8 @@ uv run playwright install
 
 ### 2. Configure LLM (Choose One)
 
+**Important:** Extraction quality varies by LLM - stronger models find more specific tips!
+
 #### Option A: OpenRouter (Recommended - Free Model!)
 ```bash
 cp .env.example .env
@@ -63,23 +65,29 @@ Extract specific optimization tips for AI services:
 scapo scrape discover --update
 
 # Step 2: Extract tips for specific services
-scapo scrape targeted --service "Eleven Labs" --limit 20
-scapo scrape targeted --service "GitHub Copilot" --limit 20
+scapo scrape targeted --service "Eleven Labs" --limit 20 --query-limit 20
+scapo scrape targeted --service "GitHub Copilot" --limit 20 --query-limit 20
 
 # Or batch process by category
-scapo scrape batch --category video --limit 15
+scapo scrape batch --category video --limit 20 --batch-size 3
 
 # Process ALL priority services one by one
-scapo scrape all --priority ultra --limit 20    # Process all ultra priority services
-scapo scrape all --dry-run                      # Preview what will be processed
+scapo scrape all --limit 20 --query-limit 20 --priority ultra    # Process all ultra priority services
+scapo scrape all --dry-run                                       # Preview what will be processed
 ```
 
 ### Key Commands:
 - `discover --update` - Find services from GitHub Awesome lists  
 - `targeted --service NAME` - Extract tips for one service
-- `batch --category TYPE` - Process multiple services (limited)
+- `batch --category TYPE` - Process ALL services in category (in batches)
 - `all --priority LEVEL` - Process ALL services one by one
 
+### Important Parameters:
+- **--query-limit**: Number of search patterns (5 = quick, 20 = comprehensive)
+- **--batch-size**: Services to process in parallel (3 = default balance)
+- **--limit**: Posts per search (20+ recommended for best results)
+
+
 ## 📚 Approach 2: Legacy Sources
 
 Use predefined sources from `sources.yaml`:
@@ -109,6 +117,27 @@ scapo models search "copilot"         # Search for specific models
 cat models/audio/eleven-labs/cost_optimization.md
 ```
 
+### 5. (Optional) Use with Claude Desktop
+
+Add SCAPO as an MCP server to query your extracted tips (from models/ folder) directly in Claude:
+
+```json
+// Add to claude_desktop_config.json
+{
+  "mcpServers": {
+    "scapo": {
+      "command": "npx",
+      "args": ["@scapo/mcp-server"],
+      "env": {
+        "SCAPO_MODELS_PATH": "path/to/scapo/models"
+      }
+    }
+  }
+}
+```
+
+Then ask Claude: "Get best practices for Midjourney" - no Python needed!
+
 ## 📊 Understanding the Output
 
 SCAPO creates organized documentation:
@@ -126,13 +155,13 @@ models/
 
 ```bash
 # ❌ Too few posts = no useful tips found
-scapo scrape targeted --service "HeyGen" --limit 5     # ~20% success rate
+scapo scrape targeted --service "HeyGen" --limit 5 --query-limit 5     # ~20% success rate
 
 # ✅ Sweet spot = reliable extraction  
-scapo scrape targeted --service "HeyGen" --limit 20    # ~80% success rate
+scapo scrape targeted --service "HeyGen" --limit 20 --query-limit 20    # ~80% success rate
 
 # 🎯 Maximum insights = comprehensive coverage
-scapo scrape targeted --service "HeyGen" --limit 30    # Finds rare edge cases
+scapo scrape targeted --service "HeyGen" --limit 30 --query-limit 20    # Finds rare edge cases
 ```
 **Why it matters:** LLMs need multiple examples to identify patterns. More posts = higher chance of finding specific pricing, bugs, and workarounds.
 
@@ -148,7 +177,7 @@ LLM_QUALITY_THRESHOLD=0.4   # More tips (less strict)
 ### "No tips extracted"
 ```bash
 # Solution: Use more posts
-scapo scrape targeted --service "Service Name" --limit 25
+scapo scrape targeted --service "Service Name" --limit 25 --query-limit 20
 ```
 
 ### "Service not found"

diff --git a/README.md b/README.md
@@ -53,15 +53,15 @@ scapo scrape discover --update
 Extract optimization tips for specific services
 
 ```bash
-scapo scrape targeted --service "Eleven Labs" --limit 20
+scapo scrape targeted --service "Eleven Labs" --limit 20 --query-limit 20
 ```
 ![Scapo Discover](assets/scrape-targeted.gif)
 
 
 Batch process multiple priority services (Recommended)
 
 ```bash
-scapo scrape batch --max-services 3 --category audio
+scapo scrape batch --category audio --batch-size 3 --limit 20
 ```
 ![Scapo Discover](assets/scrape-batch.gif)
 
@@ -89,6 +89,8 @@ uv run playwright install  # Browser automation
 
 ### 2. Configure Your LLM Provider
 
+**Note:** Extraction quality depends on your chosen LLM - experiment with different models for best results!
+
 #### Recommended: OpenRouter (Cloud)
 ```bash
 cp .env.example .env
@@ -111,14 +113,14 @@ Get your API key from [openrouter.ai](https://openrouter.ai/)
 scapo scrape discover --update
 
 # Step 2: Extract optimization tips for services
-scapo scrape targeted --service "HeyGen" --limit 20
-scapo scrape targeted --service "Midjourney" --limit 20
+scapo scrape targeted --service "HeyGen" --limit 20 --query-limit 20
+scapo scrape targeted --service "Midjourney" --limit 20 --query-limit 20
 
 # Or batch process multiple services
-scapo scrape batch --category video --limit 15
+scapo scrape batch --category video --limit 20 --batch-size 3
 
 # Process ALL priority services one by one (i.e. all services with 'ultra' tag, see targted_search_generator.py)
-scapo scrape all --priority ultra --limit 20
+scapo scrape all --limit 20 --query-limit 20 --priority ultra
 ```
 
 #### Option B: Legacy method: using sources.yaml file
@@ -196,13 +198,13 @@ scapo scrape discover --show-all        # List all services
 scapo scrape targeted \
   --service "Eleven Labs" \              # Service name (handles variations, you can put whatever --> if we don't get hit in services.json, then it will be created under 'general' folder)
   --limit 20 \                          # Posts per search (15-20 recommended)
-  --max-queries 10                      # Number of searches
+  --query-limit 20                      # Query patterns per service (20 = all)
 
 # Batch process
 scapo scrape batch \
   --category audio \                    # Filter by category
-  --max-services 3 \                    # Services to process
-  --limit 15                           # Posts per search
+  --batch-size 3 \                      # Services per batch
+  --limit 20                           # Posts per search
 
 
 ### Legacy Sources Mode
@@ -249,13 +251,52 @@ SCRAPING_DELAY_SECONDS=2                # Be respectful
 MAX_POSTS_PER_SCRAPE=100               # Limit per source
 ```
 
-### Why --limit Matters (More Posts = Better Tips)
+### Key Parameters Explained
+
+**--query-limit** (How many search patterns per service)
+```bash
+--query-limit 5   # Quick scan: 1 pattern per category (cost, optimization, technical, workarounds, bugs)
+--query-limit 20  # Full scan: All 4 patterns per category (default, most comprehensive)
+```
+
+**--batch-size** (For `batch` command: services processed in parallel)
+```bash
+--batch-size 1  # Sequential (slowest, least resource intensive)
+--batch-size 3  # Default (good balance)
+--batch-size 5  # Faster (more resource intensive)
+```
+
+**--limit** (Posts per search - More = Better extraction)
 ```bash
 --limit 5   # ❌ Often finds nothing (too few samples)
 --limit 15  # ✅ Good baseline (finds common issues)  
 --limit 25  # 🎯 Will find something (as long as there is active discussion on it)
 ```
-so, hand-wavy breakdown: With 5 posts, extraction success ~20%. With 20+ posts, success jumps to ~80%.
+Hand-wavy breakdown: With 5 posts, extraction success ~20%. With 20+ posts, success jumps to ~80%.
+
+## 🤖 MCP Server for Claude Desktop
+
+Query your extracted tips directly in Claude (reads from models/ folder - run scrapers first!):
+
+```json
+// Add to %APPDATA%\Claude\claude_desktop_config.json (Windows)
+// or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)
+{
+  "mcpServers": {
+    "scapo": {
+      "command": "npx",
+      "args": ["@scapo/mcp-server"],
+      "env": {
+        "SCAPO_MODELS_PATH": "C:\\path\\to\\scapo\\models"  // Your models folder
+      }
+    }
+  }
+}
+```
+
+Then ask Claude: "Get me best practices for GitHub Copilot" or "What models are good for coding?"
+
+See [mcp/README.md](mcp/README.md) for full setup and available commands.
 
 ## 🎨 Interactive TUI
 

diff --git a/assets/intro.gif b/assets/intro.gif
diff --git a/assets/legacy.gif b/assets/legacy.gif
diff --git a/assets/scrape-batch.gif b/assets/scrape-batch.gif
diff --git a/assets/scrape-discovery.gif b/assets/scrape-discovery.gif
diff --git a/assets/scrape-targeted.gif b/assets/scrape-targeted.gif
diff --git a/assets/tui.gif b/assets/tui.gif
diff --git a/models/audio/eleven-labs/cost_optimization.md b/models/audio/eleven-labs/cost_optimization.md
@@ -1,10 +1,25 @@
 # Eleven Labs - Cost Optimization Guide
 
-*Last updated: 2025-08-14*
+*Last updated: 2025-08-16*
 
 ## Cost & Pricing Information
 
-- 60% of credits left (~400,000 credits)
-- Subscription renewal failed due to paywall issues
+- Free trial limited to 10,000 characters per month
+- 60% of credits left (about 400,000 credits)
+- $15k saved in ElevenLabs fees
+- Free access limited to 15 minutes of voice recording per day
+- Last year I was paying +$1000/month for AI voiceovers for only one channel.
+- $29/month for unlimited usage on ElevenReader.
 - $99/month plan
+- $29/month for unlimited
+- Credits should last until June 5th
+- 10,000 free credits per month on the free plan.
+
+## Money-Saving Tips
+
+- I built my own tool, just for me. No subscriptions, no limits, just fast, clean voice generation. Cost me ~ $4/month to run.
+- MiniMax have daily credit refresh in TTS not like ElevenLabs where you need to wait 1 month to refresh.
+- Use the free plan to get 10,000 credits per month for free.
+- So, when I do, I use a temporary email to create a new account so the 10,000 chatacter limit 'resets.'
+- When converting text to voice, adding periods between letters (e.g., B.O.D.) can force the model to pronounce acronyms letter by letter, though it may consume more credits.
 
diff --git a/models/audio/eleven-labs/metadata.json b/models/audio/eleven-labs/metadata.json
@@ -1,13 +1,13 @@
 {
   "service": "Eleven Labs",
   "category": "audio",
-  "last_updated": "2025-08-14T18:53:47.086694",
+  "last_updated": "2025-08-16T13:46:28.510586",
   "extraction_timestamp": null,
   "data_sources": [
     "Reddit API",
     "Community discussions"
   ],
-  "posts_analyzed": 79,
+  "posts_analyzed": 338,
   "confidence": "medium",
   "version": "1.0.0"
 }
diff --git a/models/audio/eleven-labs/parameters.json b/models/audio/eleven-labs/parameters.json
@@ -1,6 +1,6 @@
 {
   "service": "Eleven Labs",
-  "last_updated": "2025-08-14T18:53:46.993256",
+  "last_updated": "2025-08-16T13:46:28.342822",
   "recommended_settings": {
     "setting_0": {
       "description": "voice_name=Mun W"
@@ -16,9 +16,12 @@
     }
   },
   "cost_optimization": {
-    "tip_0": "60% of credits left (~400,000 credits)",
-    "tip_1": "Subscription renewal failed due to paywall issues",
-    "pricing": "$99/month plan"
+    "tip_0": "Free trial limited to 10,000 characters per month",
+    "tip_1": "60% of credits left (about 400,000 credits)",
+    "pricing": "$29/month for unlimited",
+    "tip_3": "Free access limited to 15 minutes of voice recording per day",
+    "tip_4": "Credits should last until June 5th",
+    "tip_5": "10,000 free credits per month on the free plan."
   },
   "sources": [
     "Reddit community",

diff --git a/models/audio/eleven-labs/pitfalls.md b/models/audio/eleven-labs/pitfalls.md
@@ -1,16 +1,35 @@
 # Eleven Labs - Common Pitfalls & Issues
 
-*Last updated: 2025-08-14*
+*Last updated: 2025-08-16*
 
 ## Technical Issues
 
-### ⚠️ Unable to switch back to a Custom LLM after testing with a built-in model (gemini-2.0-flash); interface shows 'Fix the errors to proceed' even though Server URL, Model ID, and API Key are correctly filled.
+### ⚠️ Cannot switch back to a Custom LLM after testing with a built-in model (gemini-2.0-flash) on the ElevenLabs Conversational AI dashboard; even after correctly filling out Server URL, Model ID, and API Key, the interface still shows the message: 'Fix the errors to proceed' even though there is no error.
 **Fix**: Store API keys in environment variables or use a secrets manager.
 
-### ⚠️ audio plays back a female voice regardless of which option is selected when using elevenLabs API
+### ⚠️ ElevenLabs API always returns a female voice regardless of the selected gender option
+
+### ⚠️ Tasker Action Error: 'HTTP Request' (step 11) Task: 'Text To Speech To File Elevenlabs {"detail":{"status":"invalid_uid","message". "An invalid ID has been received: %voice_id'. Make sure to provide a correct one."}
 
 ## Policy & Account Issues
 
-### ⚠️ Account credits wiped (about 400,000 credits) after attempting to renew a $99/month subscription; paywall prevented payment and support ticket received no response.
+### ⚠️ Eleven Labs wiped 400,000 credits from a user's account on the $99/month plan; the user had 60% of credits left (about 400,000 credits) and was unable to renew subscription due to paywall issues.
+**Note**: Be aware of terms of service regarding account creation.
+
+### ⚠️ Free trial for ElevenLabs is limited to 10,000 characters a month, which is insufficient for scripts that are often ~20-40,000 characters long.
 **Note**: Be aware of terms of service regarding account creation.
 
+## Cost & Limits
+
+### 💰 ElevenReader credit system is considered bad by some users, making it off-putting for average consumers.
+
+### 💰 Free access to ElevenLabs is limited to 15 minutes of voice recording per day.
+
+### 💰 Free trial limited to 10,000 characters per month
+
+### 💰 Free access limited to 15 minutes of voice recording per day
+
+### 💰 $29/month for unlimited usage on ElevenReader.
+
+### 💰 $29/month for unlimited
+
diff --git a/models/audio/eleven-labs/prompting.md b/models/audio/eleven-labs/prompting.md
@@ -1,11 +1,21 @@
 # Eleven Labs Prompting Guide
 
-*Last updated: 2025-08-14*
+*Last updated: 2025-08-16*
 
 ## Tips & Techniques
 
+- I built my own tool, just for me. No subscriptions, no limits, just fast, clean voice generation. Cost me ~ $4/month to run.
+- Use ElevenLabsService(voice_name="Mun W") in Manim Voiceover
+- MiniMax have daily credit refresh in TTS not like ElevenLabs where you need to wait 1 month to refresh.
+- The ElevenLabs voice agent is the entry point into the whole system, and then it will pass off web development or web design requests over to n8n agents via a webhook in order to actually do the work.
+- Use the free plan to get 10,000 credits per month for free.
+- So, when I do, I use a temporary email to create a new account so the 10,000 chatacter limit 'resets.'
 - self.set_speech_service(ElevenLabsService(voice_name="Mun W"))
+- MacWhisper 11.10 supports ElevenLabs Scribe for cloud transcription.
 - from manim_voiceover.services.elevenlabs import ElevenLabsService
+- I built my own tool to avoid ElevenLabs fees.
+- When converting text to voice, adding periods between letters (e.g., B.O.D.) can force the model to pronounce acronyms letter by letter, though it may consume more credits.
+- ElevenLabs Scribe v1 achieves 15.0% WER on 5-10 minute patient-doctor chats, averaging 36 seconds per file.
 
 ## Recommended Settings
 

diff --git a/models/audio/firefliesai/metadata.json b/models/audio/firefliesai/metadata.json
@@ -0,0 +1,13 @@
+{
+  "service": "Fireflies.ai",
+  "category": "audio",
+  "last_updated": "2025-08-16T13:46:29.623761",
+  "extraction_timestamp": "2025-08-16T13:29:54.297790",
+  "data_sources": [
+    "Reddit API",
+    "Community discussions"
+  ],
+  "posts_analyzed": 171,
+  "confidence": "medium",
+  "version": "1.0.0"
+}
diff --git a/models/audio/firefliesai/pitfalls.md b/models/audio/firefliesai/pitfalls.md
@@ -0,0 +1,8 @@
+# Fireflies.ai - Common Pitfalls & Issues
+
+*Last updated: 2025-08-16*
+
+## Technical Issues
+
+### ⚠️ Failed to create a send channel message in Slack. Error from Slack: invalid_thread_ts
+
diff --git a/models/audio/firefliesai/prompting.md b/models/audio/firefliesai/prompting.md
@@ -0,0 +1,14 @@
+# Fireflies.ai Prompting Guide
+
+*Last updated: 2025-08-16*
+
+## Tips & Techniques
+
+- Configure Zapier to send transcripts to a channel without duplicate notifications by adjusting thread settings
+- Use custom prompts called 'apps' in Fireflies.ai to create reusable ready‑made prompts.
+- Use Zapier to send Fireflies.ai transcripts to Slack
+
+## Sources
+
+- Reddit community discussions
+- User-reported experiences