VS Code Extension
Native VS Code integration with a scan dashboard, tool status tree view, and Copilot Chat participant.
Table of contents
Overview
The AgentSec VS Code extension provides a native IDE experience for security scanning:
- Activity Bar panel with scan dashboard, tool status, and results views
- Context menu integration — right-click folders or files to scan
- Chat Participant — use
@agentseccommands in Copilot Chat - SCM integration — scan changed files from the Source Control panel
Commands
Open the Command Palette (Ctrl+Shift+P) and search for “AgentSec”:
| Command | Description |
|---|---|
| AgentSec: Scan Workspace for Security Issues | Run a full security scan on the workspace |
| AgentSec: Scan Folder for Security Issues | Scan a specific folder |
| AgentSec: Scan File for Security Issues | Scan a single file |
| AgentSec: Show Scan Dashboard | Open the scan dashboard webview |
| AgentSec: Cancel Running Scan | Abort the current scan |
| AgentSec: Refresh Tool Status | Re-detect installed security tools |
Chat Participant
The extension registers a @agentsec chat participant in Copilot Chat:
| Command | Description |
|---|---|
@agentsec /scan |
Run a full parallel security scan |
@agentsec /quick-scan |
Quick scan — skip LLM deep analysis |
@agentsec /supply-chain |
Scan dependencies for supply chain attacks |
@agentsec /results |
Show latest scan results |
@agentsec /tools |
Show available and missing security tools |
Activity Bar
The extension adds an AgentSec panel to the Activity Bar (shield icon) with three views:
Scan Dashboard
A webview that shows:
- Current scan progress and status
- Findings summary with severity breakdown
- Scan history
Security Scanners
A tree view listing all supported security tools and their installation status:
- ✅ Installed and available
- ❌ Not found — with install instructions
Use the refresh button to re-detect tools.
Scan Results
A tree view showing findings from the latest scan, organized by severity and file.
Context Menu
Right-click in the Explorer:
- On a folder → AgentSec: Scan Folder for Security Issues
- On a file → AgentSec: Scan File for Security Issues
Right-click in Source Control (changed files):
- AgentSec: Scan for Security Issues — scans only the changed files
Extension Settings
Configure the extension in VS Code Settings (Ctrl+,) under AgentSec:
| Setting | Default | Description |
|---|---|---|
agentsec.pythonPath |
python3 |
Path to Python interpreter with agentsec-core installed |
agentsec.model |
gpt-5 |
Global LLM model for all scan phases |
agentsec.modelScanners |
(empty) | Model override for Phase 2 (scanner sub-agents) |
agentsec.modelAnalysis |
(empty) | Model override for Phase 3 (LLM deep analysis) |
agentsec.modelSynthesis |
(empty) | Model override for Phase 4 (report synthesis) |
agentsec.maxConcurrent |
3 |
Max parallel scanner sessions (1–10) |
agentsec.enableLlmAnalysis |
true |
Enable LLM deep analysis phase |
agentsec.scanTimeout |
1800 |
Safety timeout in seconds |
agentsec.scanMode |
parallel |
parallel or serial execution mode |
agentsec.promptScannerSelection |
false |
Show scanner picker before each scan |
agentsec.systemMessage |
(empty) | Custom system message override |
agentsec.systemMessageFile |
(empty) | Path to system message file |
agentsec.initialPrompt |
(empty) | Custom initial prompt template |
agentsec.initialPromptFile |
(empty) | Path to initial prompt file |
agentsec.configFile |
(empty) | Path to YAML config file |
agentsec.skipScanners |
[] |
Scanner names to always exclude |
agentsec.verbose |
false |
Enable debug logging |
Leave model-specific settings empty to use the global
agentsec.modelvalue. Set them to use different models for different scan phases — e.g., a fast model for scanners and a powerful model for analysis.