Repository navigation
3.0.0
PT Application Inspector 3.0.0 - security for the agentic coding
PT Application Inspector 3.0.0 brings AI agents into the security workflow: they can inspect scan results, trigger scans, help fix vulnerabilities, and automatically check changed code before completing their work. This release also adds support for PT AI Enterprise Edition API 6.3.0 and findings from the Malicious code module.
The first standalone PT Application Inspector MCP server release is available too:
@posidev-community/ptai-mcp
What’s new
AI-powered vulnerability fixes
The new How to Fix experience generates recommendations for detected vulnerabilities and integrates directly with Quick Fix. Choose an OpenAI-compatible provider, a private proxy, or a locally deployed model, review the suggested change, apply it, or request another option.
MCP: connect your coding agent to application security
The plugin now integrates with external AI agents through the Model Context Protocol (MCP). Agents such as Cursor, Claude, and Codex can work with scan results directly in the IDE:
- start local scans and check their status;
- list vulnerabilities and retrieve detailed findings;
- inspect data-flow and exploitation information;
- confirm, discard, or triage vulnerabilities;
- suggest code changes based on the actual findings.
Every action is reflected immediately in the PT Application Inspector UI, so the developer and the agent always see the same security state.
To connect an installed agent, run PT Application Inspector: Connect to AI Agent. For terminal and headless workflows, install the standalone MCP server:
npx -y @posidev-community/ptai-mcp installThe standalone package requires Node.js 18 or later.
GenAI Code Gates: check generated code before it ships
AI hooks add a security gate to the agent loop. They are configured globally for each selected AI agent and become available across that agent's projects. At the start of a turn, the integration records a baseline; when the agent finishes, an incremental local scan is triggered only if project files have changed. If nothing changed, the scan is skipped.
Automatic checks run only in projects where PT Application Inspector has already been used and at least one scan has completed. Hooks never start the first scan in a new project - you remain in control of when analysis is initialized. Once a scan is available, the agent receives a summary of newly detected vulnerabilities and can iterate through generated → checked → fixed until the code is ready.
You can opt out for an individual project via PT Application Inspector: Disable AI Hooks for This Project. If you change your mind, run PT Application Inspector: Enable AI Hooks for This Project to remove that project-level restriction. Loop-protection mechanisms prevent repeated scans within the same agent turn.
More coverage and current server compatibility
- Added display and triage support for vulnerabilities detected by the Malicious code module.
- Added full support for PT AI Enterprise Edition API 6.3.0.
Getting started
- Install or update the PT Application Inspector extension.
- Make sure the PT Application Inspector analyzer is installed. See the installation instructions.
- Open a project and run a scan.
- For AI workflows, run PT Application Inspector: Connect to AI Agent, then enable MCP and (optionally) AI hooks for the agents you use.
Requirements
- Visual Studio Code 1.72.0 or later
- 8 GB RAM
- 5 GB of free disk space
- A supported 64-bit operating system (Windows, Linux, or macOS; see the full requirements)
Upgrade
This is a major release. After updating, review the new How to Fix provider settings and connect your AI agents if needed. AI hooks are enabled globally for the agents you select; use the command palette to disable or re-enable them for an individual project.
Download: Open VSX Registry · VS Code Marketplace · Standalone MCP package


