Fix bugs in one shot with AI-powered context gathering
DebugContext is an MCP Server that works as a plugin for AI coding tools like Claude Code, Cursor, and Windsurf. When a developer encounters a bug, it automatically extracts file paths and line numbers from error messages, reads the relevant source code, collects Git change context, and generates a structured analysis report — so the AI can fix the bug on the first try.
Before: User pastes error → AI guesses → wrong fix → guesses again → wastes tokens
After: User pastes error → DebugContext gathers context → AI fixes it correctly in one shot
npm install -g debugctxThat's it. Auto-detects your AI tools, configures MCP globally. Restart your AI tool and it works in ALL projects.
Or run per-project:
npx debugctx init| Tool | Supported |
|---|---|
| Claude Code | ✓ |
| Cursor | ✓ |
| Windsurf | ✓ |
| Cline | ✓ |
| Continue | ✓ |
When the AI encounters an error, it automatically calls the analyze_bug tool:
- Error parsing — Regex matches error formats from 14 programming languages, extracts file paths and line numbers
- Code reading — Automatically reads code around the error line (N lines above and below)
- Git collection — Collects the last 8 commits, unstaged changes, and the latest commit diff
- Project scanning — Scans the project file structure to help AI understand code organization
- Missing context detection — Detects whether context is sufficient, tells AI what to ask the user
- Token statistics — Shows analysis time, report size, and estimated token consumption at the end
| Language | Example Format |
|---|---|
| JavaScript / TypeScript | at func (file.js:42:10) |
| JSX / TSX / Vue | at Component (file.jsx:42:10) |
| Python | File "file.py", line 42 |
| PHP | in /path/file.php on line 42 |
| Go | file.go:42 |
| Ruby | file.rb:42 |
| Java | at com.example.File.java:42 |
| Rust | at file.rs:42:10 |
| CSS / SCSS | file.css:42 |
| HTML | file.html:42 |
Automatically called by the AI when a user encounters any bug, error, crash, or unexpected behavior.
| Parameter | Type | Required | Description |
|---|---|---|---|
error_message | string | Yes | Error message, stacktrace, or problem description |
related_files | string[] | No | User-mentioned related file paths |
project_dir | string | No | Project root directory (defaults to cwd) |
context_lines | number | No | Lines of context above/below error (default 40) |
Search project source code by keyword to find function definitions, class references, variable locations, etc.
| Parameter | Type | Required | Description |
|---|---|---|---|
keyword | string | Yes | Search keyword |
file_type | string | No | Limit to file extension (e.g. php, js) |
project_dir | string | No | Project root directory (defaults to cwd) |
The tool auto-detects system language, supporting Chinese and English.
npx debugctx init # Auto-detect system language
npx debugctx init --lang zh # Force Chinese
npx debugctx init --lang en # Force EnglishDetection priority: --lang flag > system env vars > default English
If auto-detection doesn't work, add the following to your AI tool's MCP config:
{
"mcpServers": {
"debugctx": {
"command": "node",
"args": ["/absolute/path/to/debugctx/src/server.js"]
}
}
}Config file locations (global install writes here):
| Tool | Config File |
|---|---|
| Claude Code | ~/.claude/mcp.json |
| Cursor | ~/.cursor/mcp.json |
| Windsurf | ~/.codeium/mcp.json |
| Cline | ~/.cline/mcp.json |
| Continue | ~/.continue/debugctx.json |
- Node.js >= 18
- Windows / macOS / Linux