Agentic Debugger Bridge – AI-Powered Visual Studio Control

FeaturesHow It WorksAPISecurityGet Started
Visual Studio Extension

Give AI Agents Control of Your Debugger

The Agentic Debugger Bridge exposes Visual Studio’s debugger, build system, and code analysis via HTTP — so AI agents can see, control, and understand your code in real time.

View on GitHubQuick Start
<100ms
WebSocket Latency
20+
API Endpoints
10x
Faster with Batch Commands

Everything Your AI Agent Needs

One extension. Full IDE control. Zero UI automation hacks.

🔎

Roslyn Code Analysis

Search symbols across your entire solution. Go to definition, find all references, get document outlines, and retrieve type information — all via API.

🐧

Full Debugger Control

Start/stop debugging, step through code, set breakpoints, evaluate expressions, inspect locals and call stacks — all programmatically.

Real-Time WebSocket

Push notifications for debugger state changes with sub-100ms latency. No more polling. Your agent reacts instantly to breakpoint hits, build completions, and errors.

🔧

Build System Integration

Trigger builds, rebuilds, and clean operations. Read compilation errors as structured JSON. Get build output logs programmatically.

📊

Observability & Metrics

Real-time performance metrics, health monitoring, complete request audit trail. Know exactly what your agent is doing and how fast it’s doing it.

🌐

Multi-Instance Routing

Run multiple Visual Studio instances. The primary bridge auto-discovers secondary instances and routes commands to the right one. One API, multiple IDEs.

How It Works

Install once. Your agent discovers and connects automatically.

1

Install the Extension

Install the VSIX in Visual Studio 2022. The bridge starts automatically when VS opens. Port 27183, zero configuration.

2

Agent Discovers Bridge

Your agent reads %TEMP%\agentic_debugger.json for port and connection details. No hardcoded config needed.

3

Control Everything

HTTP commands to debug, build, and analyze code. WebSocket for real-time state. Your agent has full IDE awareness.

Clean, Simple API

Standard HTTP. OpenAPI documented. Works with any language.

GET /stateGet current debugger state
curl -H "X-Api-Key: dev" http://localhost:27183/state

{
  "mode": "Break",
  "isDebugging": true,
  "currentLine": 42,
  "currentFile": "Program.cs",
  "locals": [
    { "name": "userId", "value": "123", "type": "int" }
  ]
}
POST /commandExecute debugger actions
curl -X POST http://localhost:27183/command \
  -H "X-Api-Key: dev" \
  -H "Content-Type: application/json" \
  -d '{"action": "SetBreakpoint", "file": "Program.cs", "line": 42}'

{ "ok": true, "message": "Breakpoint set at Program.cs:42" }
POST /code/symbolsRoslyn semantic search
curl -X POST http://localhost:27183/code/symbols \
  -H "X-Api-Key: dev" \
  -H "Content-Type: application/json" \
  -d '{"query": "UserService", "kind": "class"}'

{
  "results": [
    { "name": "UserService", "file": "Services/UserService.cs",
      "line": 15, "kind": "class" }
  ]
}

Full API docs available at http://localhost:27183/docs when the extension is running, or via /swagger.json for OpenAPI 3.0 spec.

Security First

Granular permissions. Read-only by default. You control what your agent can do.

Enabled by Default (Safe)

  • ✓ Code Analysis — symbol search, go-to-definition, references
  • ✓ Observability — debugger state, metrics, errors, logs

Opt-In (Requires Enabling)

  • ⚠ Debug Control — start/stop, step, execution
  • ⚠ Build System — build, rebuild, clean
  • ⚠ Breakpoints — set and clear
  • ⚠ Configuration — settings, expressions, watches

Configure permissions in Tools > Options > Agentic Debugger > Permissions

Use Cases

From autonomous debugging to intelligent code review.

AI Debugging Agent

Your agent sets a breakpoint, runs the app, inspects locals when it breaks, evaluates expressions, and fixes the bug — without touching the UI.

Build Monitor

Hook into WebSocket events. Get instant notification when builds fail. Parse structured error output. Trigger automated fixes.

Code Intelligence

Use Roslyn to navigate any .NET codebase semantically. Find symbol definitions, trace references, understand type hierarchies — all via HTTP.

Get Started in 60 Seconds

Three steps to give your AI agent full IDE awareness.

Step 1: Install the Extension

Download the latest VSIX from the GitHub releases page and double-click to install.

AgenticDebugger-v2.1.vsix  →  double-click to install

Step 2: Open Visual Studio

The bridge starts automatically. Verify it’s running:

curl http://localhost:27183/status
# {"version":"2.1","permissions":{...}}

Step 3: Connect Your Agent

Use any HTTP client. Here’s a Python example:

import requests

base = "http://localhost:27183"
headers = {"X-Api-Key": "dev"}

# Get debugger state
state = requests.get(f"{base}/state", headers=headers).json()

# Search for a class
symbols = requests.post(f"{base}/code/symbols",
    headers=headers,
    json={"query": "UserService", "kind": "class"}).json()

# Set a breakpoint
requests.post(f"{base}/command",
    headers=headers,
    json={"action": "SetBreakpoint",
          "file": "Program.cs", "line": 42})

Open Source. Free Forever.

Built for the agentic development community. MIT licensed.

GitHub RepositoryExample Agents

Agentic Debugger Bridge — Built by Martien de Jong

ENNL