Activity Feed
The Activity Feed shows everything Claude Code is doing in real time. It displays reasoning, file operations, commands, and any blocked actions.
Feed Structure
The activity feed shows events chronologically:
[11:23:45] Claude: Analyzing the request...[11:23:47] Claude: I'll create a Python script to process the data.
Plan:1. Load data from input.csv2. Apply transformations3. Save results to output.csv
Files to create: process_data.pyCommands to run: python3 process_data.py input.csv
[Waiting for approval]
[11:24:02] ✅ Approved by user[11:24:03] ✅ Write: process_data.py[11:24:05] 🖥️ Run: python3 process_data.py input.csv[11:24:07] Output: Processed 1,247 rows. Saved to output.csv
Complete · Free (Max) · 22.3sEvent Types
Reasoning
Claude’s thinking process is shown as plain text:
Claude: Looking at the error message, the issue is a missingimport statement. The script tries to use pandas but doesn'timport it at the top.Plan Phase
During planning, Claude shows what it intends to do:
Plan:1. Add import pandas at the top of script.py2. Run the script again to verify
Files to modify: script.pyCommands to run: python3 script.pyFile Operations
File operations show success (✅) or failure (❌):
✅ Write: analysis/results.csv✅ Edit: scripts/process.py (lines 45-52)✅ Read: config/settings.yamlBash Commands
Commands show the command and output:
🖥️ Run: python3 analyze.py --input data.csv
Output:Loading data... doneProcessing 5,000 samplesAccuracy: 94.2%Saved model to models/classifier.pklBlocked Actions
When permission profiles block an action:
🚫 Blocked: rm -rf ~/important_data Reason: Command not in allow list
[Approve and Re-run] button availableErrors
Errors are shown with details:
❌ Error running: python3 train.py
Traceback (most recent call last): File "train.py", line 15, in <module> import torchModuleNotFoundError: No module named 'torch'Status Indicators
At the bottom of the activity feed:
| Indicator | Meaning |
|---|---|
| Complete | Task finished successfully |
| Failed | Task encountered an error |
| Waiting | Plan ready for approval |
| Running… | Execution in progress |
Cost Display
The cost section shows:
Complete · Free (Max) · 23.4s- Free (Max): Using Claude Max subscription (no API cost)
- $0.12 (API): If using API key, shows the cost
- 23.4s: Total execution time
Scrolling Behavior
- Feed auto-scrolls as new events arrive
- Click anywhere to pause auto-scroll
- Scroll to bottom to resume auto-scroll
Filtering (Coming Soon)
Future versions will support filtering by:
- Event type (reasoning, files, commands)
- Success/failure
- Time range
Copying Output
Right-click on any output block to copy its contents. Useful for:
- Sharing error messages
- Saving command output
- Debugging issues
Best Practices
- Watch the plan — Understand what Claude intends before approving
- Check file paths — Verify files are written where expected
- Read errors carefully — Error messages often suggest the fix
- Export history — Use gear menu → Export History for a permanent record
Next Steps
- Quick Drop Mode — For quick file-based tasks
- Session Continuity — Build on previous work