You don't need to understand the internals to use VIBES. These tools handle the hard parts — validating audit data, generating reports, scoring risk, and tracking agent decisions — so you can focus on building.
vibecheck is the main command-line tool for working with VIBES data. It validates your audit trail, generates readable reports, and scores risk — for both AI-generated code and agent decisions. Think of it as the dashboard for your AI flight recorder.
Check that your .ai-audit/ data is correct and complete. Catches missing records, broken references, and schema issues — whether it's code annotations or agent decision logs.
Generate visual reports of everything AI did in your project. See which models wrote code, which agents made decisions, coverage stats, and trends over time — in HTML or JSON.
Compute PRISM risk scores for every AI action. Flag high-risk code changes and critical agent decisions automatically. Use it in CI/CD to gate deployments.
Each command works across all your VIBES data — code annotations, agent decision records, delegation chains, and more.
npx vibecheck init
Set up VIBES in your project. Creates the .ai-audit/ folder and configures your assurance level. Takes about 30 seconds.
npx vibecheck verify
Validate all your audit data. Checks that every AI-generated code annotation and agent decision record is complete, properly formatted, and internally consistent.
npx vibecheck report
Generate a summary of all AI involvement in your project. See which models touched which files, what decisions agents made, and how much of your codebase is AI-assisted.
npx vibecheck report --format html
Same report, but as a visual HTML page you can open in your browser or share with your team. Great for sprint reviews and compliance documentation.
npx vibecheck risk
Compute and display PRISM risk scores. Shows which code changes and agent actions carry the most risk, so you know what to review first.
npx vibecheck risk --threshold 0.6 --ci
Run risk scoring in CI/CD mode. Fails the pipeline if any AI action exceeds your risk threshold — whether it's a code change or an agent approving a deployment.
Install: vibecheck requires Node.js 18+. Run it directly with npx vibecheck or install globally with npm install -g vibecheck.
If you want everything in one place without stitching tools together, Maestro is the easiest way to get started. It's a full implementation of the VIBES ecosystem — audit tracking, verification, risk scoring, and agent governance — built in and ready to go.
Maestro handles the entire VIBES ecosystem. No manual instrumentation, no configuration — start working and everything gets tracked automatically. Code generation, agent decisions, delegation chains, risk scores — all of it.
Full audit trail + cryptographic signing + risk scoring + agent governance
Get started at runmaestro.ai →Zero setup. VIBES tracking starts automatically when you use AI tools through Maestro. Every line of code and every agent decision gets recorded without configuration.
Built-in verification. VERIFY signing happens in the background. Your audit data gets cryptographic proof of authenticity with no extra steps.
Automatic risk scoring. PRISM evaluates every AI action as it happens. High-risk code changes and agent decisions get flagged before they ship.
Agent governance. EVOLVE tracks what your agents decided, why, and what happened. Patterns emerge over time, making your agents smarter and more accountable.
VIBES works with the AI tools you already use. Here's the current status of native integrations — each one records both code generation and agent actions automatically.
Records code it writes, decisions it makes, and reasoning it uses. Full VIBES tracking for every Claude Code session.
Captures code generation, tool use, and agent operations. Works with Gemini CLI and API integrations.
Tracks code output, API calls, and autonomous agent actions. Integrates with Codex's sandbox execution model.
IDE-integrated VIBES tracking for Cursor sessions. Records AI edits and inline completions.
Windsurf integration is in development. Planned to capture AI-assisted edits and agent flows.
GitHub Copilot integration planned. Will track completions, chat suggestions, and workspace agent actions.
Don't see your tool? Ask your provider to add VIBES support — the more people who ask, the faster it happens.
You've seen the tools. Now pick a path and add transparency to your AI workflow today — for code generation and agent operations alike.
Pick your tool above and run the install command. You'll have a working audit trail in under 5 minutes.
Full install guide →Don't see your tool listed? Ask the provider to add native VIBES support — the more people who ask, the faster integrations ship.
How to ask →Skip the setup entirely. Maestro handles VIBES, VERIFY, PRISM, and EVOLVE with zero configuration.
runmaestro.ai →