Fix: Don't reset chat when generating new agent - Allow agent creation during ongoing conversations

PUBMED_SETUP.md

@@ -0,0 +1,217 @@
+# 🔬 PubMed Tool Configuration Guide
+
+## Environment Variable Required
+
+To enable the PubMed search tool in your ID Agents app, you need to set the following environment variable:
+
+### **`NCBI_EMAIL`**
+
+This is your email address that will be sent to NCBI (National Center for Biotechnology Information) with each API request.
+
+---
+
+## Why is Email Required?
+
+NCBI requires an email address for E-utilities API access for the following reasons:
+
+1. **API Usage Monitoring**: NCBI uses email to track API usage patterns
+2. **Contact for Issues**: If there are problems with your requests, NCBI can contact you
+3. **Terms of Service**: It's part of NCBI's E-utilities usage guidelines
+4. **Rate Limiting**: Helps NCBI manage and enforce rate limits appropriately
+
+---
+
+## How to Set It Up
+
+### Option 1: HuggingFace Space (Recommended)
+
+1. Go to your HuggingFace Space: https://huggingface.co/spaces/John-jero/IDWeekAgents
+2. Click on **"Settings"** (gear icon)
+3. Navigate to **"Repository secrets"**
+4. Click **"Add a secret"**
+5. Enter:
+   - **Name**: `NCBI_EMAIL`
+   - **Value**: Your email address (e.g., `[email protected]`)
+6. Click **"Save"**
+7. **Restart your space** for changes to take effect
+
+### Option 2: Local Development (.env file)
+
+If running locally, create/edit your `.env` file:
+
+```bash
+# .env file
+OPENAI_API_KEY=your-openai-api-key-here
+[email protected]
+```
+
+---
+
+## What Email Should I Use?
+
+### ✅ Recommended:
+- **Institutional email**: `[email protected]`
+- **Work email**: `[email protected]`
+- **Academic email**: `[email protected]`
+
+### ⚠️ Acceptable:
+- Personal email (Gmail, Yahoo, etc.)
+- Any valid email address
+
+### ❌ Not Recommended:
+- Fake or invalid emails
+- Generic emails like `[email protected]`
+- Temporary/disposable emails
+
+**Best Practice**: Use a real, professional email address associated with your institution or work.
+
+---
+
+## Optional: NCBI API Key
+
+For **higher rate limits**, you can also register for a free NCBI API key:
+
+### How to Get an NCBI API Key:
+
+1. Go to: https://www.ncbi.nlm.nih.gov/account/
+2. **Sign in** or **Register** for an NCBI account
+3. Go to **Settings** → **API Key Management**
+4. Click **"Create an API Key"**
+5. Copy your API key
+
+### Add to HuggingFace Space:
+
+- **Name**: `NCBI_API_KEY`
+- **Value**: Your NCBI API key
+
+### Benefits:
+- **Without API Key**: 3 requests/second
+- **With API Key**: 10 requests/second
+
+---
+
+## Default Behavior
+
+If you **don't set** `NCBI_EMAIL`, the tool will use a default:
+
+```python
+# From tools/pubmed_search.py line 67:
+if not email:
+    email = "[email protected]"
+    logger.info("Using default email for NCBI API access")
+```
+
+This will work, but it's **better to set your own email** for:
+- ✅ Proper attribution
+- ✅ NCBI can contact you if needed
+- ✅ Better compliance with NCBI guidelines
+
+---
+
+## How to Verify It's Working
+
+### Test in Your Agent:
+
+1. **Create an agent** with the **"search_pubmed"** skill
+2. **Chat with the agent** and ask:
+   ```
+   "Search PubMed for recent articles on C. difficile treatment"
+   ```
+3. **Check the response**:
+   - ✅ Should return recent articles with titles, authors, PMIDs
+   - ❌ If error, check your email is set correctly
+
+### Check Logs:
+
+In your HuggingFace Space logs, you should see:
+```
+INFO: Using email: [email protected] for NCBI API access
+```
+
+---
+
+## Complete Environment Variables Setup
+
+For full functionality, set these in HuggingFace Space Settings → Repository secrets:
+
+### **Required**:
+```
+OPENAI_API_KEY=sk-...your-openai-key
+AUTH_CREDENTIALS=username:password,user2:pass2
+```
+
+### **Optional (but recommended)**:
+```
+[email protected]
+NCBI_API_KEY=your-ncbi-api-key
+SERPER_API_KEY=your-serper-key
+```
+
+---
+
+## Troubleshooting
+
+### Problem: PubMed search returns no results or errors
+
+**Solution 1**: Check email is set
+```bash
+# In Space Settings → Secrets, verify:
+NCBI_EMAIL = [email protected]
+```
+
+**Solution 2**: Restart your space after adding the secret
+
+**Solution 3**: Check logs for specific error messages
+
+### Problem: Rate limit errors
+
+**Solution**: Add `NCBI_API_KEY` to increase rate limit from 3 to 10 requests/second
+
+### Problem: "Invalid email" errors
+
+**Solution**: Use a properly formatted email address: `[email protected]`
+
+---
+
+## Code Reference
+
+The PubMed tool code is in: `tools/pubmed_search.py`
+
+Key lines:
+```python
+# Line 37: Email parameter with environment variable default
+"email": {"type": "string", "description": "user email for NCBI API", 
+          "default": os.getenv("NCBI_EMAIL", "")}
+
+# Line 62-67: Email fallback logic
+if not email:
+    email = os.getenv("NCBI_EMAIL", "")
+
+if not email:
+    email = "[email protected]"
+    logger.info("Using default email for NCBI API access")
+```
+
+---
+
+## Quick Setup Checklist
+
+- [ ] Go to HuggingFace Space Settings
+- [ ] Navigate to Repository secrets
+- [ ] Add secret: `NCBI_EMAIL` = your email
+- [ ] (Optional) Add secret: `NCBI_API_KEY` = your API key
+- [ ] Restart your space
+- [ ] Test PubMed search with an agent
+- [ ] Verify results are returned
+
+---
+
+## Summary
+
+**Environment Variable**: `NCBI_EMAIL`  
+**Value**: Your email address  
+**Location**: HuggingFace Space Settings → Repository secrets  
+**Required**: Technically optional, but strongly recommended  
+**Format**: `[email protected]`
+
+**After setting it, restart your space and the PubMed tool will work with your email!** 📧🔬

app.py

@@ -2158,10 +2158,7 @@ def build_ui():
                     web_access_toggle, allow_fallback_toggle,
                     uploaded_files, link1, link2, link3, link4, challenger_toggle],
             outputs=[agent_output]       # <- only the JSON
-        ).then(                          # 2) show initial instruction instead of preload demo chat
-            lambda: show_initial_instruction_state(),
-            inputs=[], outputs=[builder_chatbot, chat_input, builder_send_button, reset_button, invocation_log, active_children]
-        ).then(                          # 2.5) auto-load agent fields into builder panel
+        ).then(                          # 2) auto-load agent fields into builder panel
             lambda agent_json: (
                 json.loads(agent_json).get("agent_type", ""),
                 json.loads(agent_json).get("agent_name", ""),