You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This project uses the following sponsor APIs and platforms:
45
-
46
-
-**Anthropic** - Claude API for MCP integration testing and AI-powered debugging capabilities
47
-
-**Gradio** - `@gradio/client` for connecting to Gradio-powered APIs in the demo app
48
-
49
-
---
50
36
51
37

52
38
53
39
## Key Features
54
40
55
41
### 🎯 Visual Context
42
+
56
43
Click any element to instantly send its source code location, computed styles, and DOM hierarchy to AI. No more explaining "it's the blue button in the header".
57
44
58
45
### 🛠️ Full DevTools Access
46
+
59
47
AI can access Chrome DevTools to analyze network requests, console logs, and performance metrics. It sees what you see.
60
48
61
49
### 🤖 Multi-Agent Workflow
50
+
62
51
Switch between agents (Claude Code, Goose) and track their debugging progress visually with step-by-step status updates.
You can customize available AI agents and set a default agent:
312
302
313
303
```typescript
@@ -339,11 +329,11 @@ export default {
339
329
```
340
330
341
331
**Key Features:**
332
+
342
333
- Custom agents with the **same name** as [default agents](https://agentclientprotocol.com/overview/agents) automatically inherit missing properties (icons, env)
343
334
- You can override just the command/args while keeping default icons
344
335
- If no custom agents provided, defaults are: Claude Code, Codex CLI, Gemini CLI, Kimi CLI, Goose, OpenCode
345
336
346
-
347
337
## What It Does
348
338
349
339
**Click element → Describe issue → AI analyzes → Get fix**
@@ -354,68 +344,81 @@ export default {
354
344
4. Get intelligent solutions through natural conversation
355
345
356
346
**Examples:**
347
+
357
348
- "Why is this button not clickable?" → AI checks `pointer-events`, z-index, overlays
358
349
- "This API call is failing" → AI analyzes network requests, timing, responses
359
350
- "Where is this component?" → Jump to source file and line number
360
351
361
352
## Two Workflow Modes
362
-
363
-
DevInspector offers two ways to interact with your AI, depending on your preference:
364
-
365
-
### 1. Editor Mode
366
-
**Best for:** Code-heavy tasks, refactoring, and maintaining flow.
367
-
368
-
-**How it works:** You use your IDE's AI assistant (Cursor, Windsurf, Copilot).
369
-
-**The Flow:** Click an element in the browser -> The context (source, props, styles) is sent to your Editor via MCP -> You ask your Editor to fix it.
370
-
-**Why:** Keeps you in your coding environment.
371
-
372
-
### 2. Inspector Bar Mode (Recommended)
373
-
**Best for:** Quick fixes, visual tweaks, or if you don't use an AI editor.
374
-
375
-
-**How it works:** You use the floating "Inspector Bar" directly in the browser.
376
-
-**The Flow:** Click "Ask AI" in the browser -> Select an agent (e.g., Claude Code, Custom Script) -> The agent runs in your terminal but interacts with the browser overlay.
377
-
-**Why:** No context switching. Great for "what is this?" questions or network debugging.
378
-
379
-
## MCP Tools
353
+
354
+
DevInspector offers two ways to interact with your AI, depending on your preference:
355
+
356
+
### 1. Editor Mode
357
+
358
+
**Best for:** Code-heavy tasks, refactoring, and maintaining flow.
359
+
360
+
-**How it works:** You use your IDE's AI assistant (Cursor, Windsurf, Copilot).
361
+
-**The Flow:** Click an element in the browser -> The context (source, props, styles) is sent to your Editor via MCP -> You ask your Editor to fix it.
362
+
-**Why:** Keeps you in your coding environment.
363
+
364
+
### 2. Inspector Bar Mode (Recommended)
365
+
366
+
**Best for:** Quick fixes, visual tweaks, or if you don't use an AI editor.
367
+
368
+
-**How it works:** You use the floating "Inspector Bar" directly in the browser.
369
+
-**The Flow:** Click "Ask AI" in the browser -> Select an agent (e.g., Claude Code, Custom Script) -> The agent runs in your terminal but interacts with the browser overlay.
370
+
-**Why:** No context switching. Great for "what is this?" questions or network debugging.
371
+
372
+
## MCP Tools
380
373
381
374
### `capture_element_context`
375
+
382
376
Activates visual selector. Returns source location, DOM hierarchy, styles, dimensions, and user notes.
383
377
384
378
### `list_inspections`
379
+
385
380
Shows all inspections with ID, element details, notes, and status (pending/in-progress/completed/failed).
386
381
387
382
### `update_inspection_status`
383
+
388
384
Updates inspection status with optional progress steps.
389
385
390
386
**Parameters:**`status`, `message` (required for completed/failed), `progress`, `inspectionId` (optional)
391
387
392
388
### `execute_page_script`
389
+
393
390
Executes JavaScript in browser context. Access to window, document, React/Vue instances, localStorage.
394
391
395
392
### `chrome_devtools`
393
+
396
394
Agentic tool for Chrome DevTools access. Provides network inspection, console logs, performance metrics, element interaction, and more.
397
395
398
396
## MCP Prompts
399
397
400
398
### `capture_element`
399
+
401
400
Capture and analyze UI element context.
402
401
403
402
### `view_inspections`
403
+
404
404
View all pending, in-progress, and completed inspections.
💡 Optional if Chrome is already open. Use when you need to launch a new Chrome instance.
412
413
413
414
### `get_network_requests`
415
+
414
416
List network requests or get details of a specific one. Always refreshes the list first.
415
417
416
418
**Parameter:**`reqid` (optional) - If provided, get details for that request. If omitted, just list all requests.
417
419
418
420
### `get_console_messages`
421
+
419
422
List console messages or get details of a specific one. Always refreshes the list first.
420
423
421
424
**Parameter:**`msgid` (optional) - If provided, get details for that message. If omitted, just list all messages.
@@ -425,6 +428,7 @@ List console messages or get details of a specific one. Always refreshes the lis
425
428
For a deep dive into how the MCP context, CMCP library, and Puppet binding mechanism work together, see the [Architecture Documentation](./docs/architecture/mcp-cmcp-puppet-architecture.md).
426
429
427
430
**Key concepts:**
431
+
428
432
-**Hub-and-spoke model**: Vite dev server acts as central hub managing multiple client connections
0 commit comments