AgentReady
A verification layer for AI-generated code

Catch risky AI code
before you commit.

AgentReady sits between your AI agent and your git history — scanning every diff for risk, verifying the agent built what you asked for, and writing the commit message before you push.

Download free — MacSee what it checks

Free forever. No credit card required.

Home

Your local readiness workflow

Open a repo, run a baseline check, revisit report history, or jump into the Pro workspace.

AgentReady home screen

Pro Workspace

Alignment Review, AgentForge, and AgentNarrator in one place

The Pro workspace keeps every premium tool together instead of buried inside the free flow.

AgentReady Pro workspace
Free — runs on every diff

Six checks that run
before every commit.

Every check scans your uncommitted diff locally — no upload, no account. It catches the mistakes AI agents make most often and generates a repair prompt you can paste straight back into Cursor or Claude.

Also includes test runner integration and saved report history.

🧪

Deleted tests

Catches removed or gutted test coverage.

🔑

Hardcoded secrets

API keys and tokens left in the diff.

📦

Dependency changes

New packages or version bumps.

🚧

Placeholder code

TODOs and stubs in production paths.

📐

Oversized diffs

Suspiciously broad changes.

🖨️

Console log leakage

Debug output left in the code.

Pro

Did the agent build what you asked?

Alignment Review adds Claude as a second opinion. It reads your diff against what you described and returns a verdict — not just whether the code is risky, but whether it's right.

Alignment verdict

Aligned or not-aligned with a confidence level.

Unrelated files

Changed files that have nothing to do with the feature.

Scope creep

Changes that go further than what you asked for.

Suggested fixes

Actionable guidance you can hand back to the agent.

Alignment Review — result
ALIGNEDhigh confidence

The diff matches the requested feature. A 404 JSON response was added for missing user IDs with appropriate test coverage.

Unrelated files (1)

⚠ apps/web/app/layout.tsx — no changes related to the 404 feature

Suggested fix

Revert layout.tsx or confirm the change was intentional before committing.

Pro — AgentForge

Agents work better when they
know your stack.

AgentForge detects your tech stack and writes .cursorrules and AGENTS.md to your repo. Cursor picks up .cursorrules automatically. Claude Code reads AGENTS.md. Both agents hit the ground running from prompt one.

Detects your framework, language, and toolchain
Writes both files in one click from the Pro workspace
Regenerate any time the stack changes
.cursorrulesgenerated by AgentForge

Stack: Next.js 15, TypeScript, Tailwind CSS

Auth: Clerk — never build custom auth

Style: Tailwind only, no inline styles

Types: Always define types, never use any

Tests: npm test — run before finishing

AGENTS.mdgenerated by AgentForge

# AgentReady — Agent Context

Framework: Next.js 15 · TypeScript · Tailwind

Test command: npm test

Do not modify /app/api/webhooks without explicit instruction.

Follow component patterns in /components.

AgentNarrator — output

Commit message

feat(api): add 404 response for missing users

Returns a JSON error body with status 404 when a
user ID is not found. Adds test coverage for the
new error path in user.controller.ts.

PR description

Add 404 response for missing user IDs

**What changed:** API now returns a structured
404 JSON error when a user is not found.
**Why:** Aligns with the API error contract and
avoids silent 200s on missing resources.

Copy commitCopy PR
Pro — AgentNarrator

The commit message
writes itself.

After every Pro check, AgentNarrator generates a commit message and full PR description from your spec and results. No rewriting what the agent did in your own words.

Copy to clipboard or commit directly from the app — AgentReady stages your files and runs the commit with the generated message.

MCP server

Run checks without
leaving your editor.

AgentReady ships an MCP server you can plug into Cursor, Claude Code, or any MCP-compatible assistant. Trigger a readiness check mid-session — verdict and repair prompt come back inline.

run_readiness_check

Run a full check on any local repo. Returns verdict, findings, and repair prompt.

get_latest_report

Retrieve the last saved report without rerunning the check.

Pro over MCP

Authenticated Pro users get alignment review in MCP responses too.

mcp.json — Cursor / Claude Code

{

"mcpServers": {

"agentready": {

"command": "npx",

"args": ["agentready-mcp"]

}

}

}

Then ask your assistant: "Run an AgentReady check on this repo."

Simple pricing

Start free. Upgrade when alignment review starts saving you time.

Free

Catch obvious risks on every diff

$0
  • Unlimited checks
  • Diff risk scan
  • Test runner integration
  • Repair prompt
  • Saved report history
  • MCP server
Get started free
Popular

Pro

Verify the agent built the right thing

$9/mo
  • ◆Everything in Free
  • ◆AI alignment review
  • ◆Unrelated file detection
  • ◆Scope creep analysis
  • ◆AgentNarrator
  • ◆AgentForge
Get started — $9/mo

Full feature breakdown on the pricing page.

Ship AI code you actually trust

Download AgentReady, open any repo, and run your first check in under a minute. Free forever — no account required to get started.

Download free — MacView pricing