> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kipmox.com/llms.txt
> Use this file to discover all available pages before exploring further.

# kipmox Agent

> Give your AI coding agent a Salesforce reviewer — deterministic analysis and independent fix verification, over MCP.

**kipmox Agent** brings kipmox's Salesforce guardrails to the AI coding agent you already use. It runs as an [MCP](https://modelcontextprotocol.io) server, so agents like **Claude Code**, **Cursor**, and **Codex** can call it directly.

kipmox Agent doesn't write your code — it **checks it**. Your agent stays in charge of the fix; kipmox is the deterministic reviewer and verifier alongside it.

<Note>
  kipmox Agent is the same analysis and verification engine as the [VS Code extension](/getting-started/introduction), delivered as a channel your AI agent can drive. The extension applies fixes itself; with kipmox Agent, **your agent writes the fix and kipmox verifies it**.
</Note>

## How it works

One prompt in your agent runs the whole loop:

<Steps>
  <Step title="analyze_code — kipmox reviews">
    kipmox runs its deterministic Salesforce rules (and your project's own ruleset, when present) over the code and returns structured findings — rule, line, message, and a remediation hint.
  </Step>

  <Step title="Your agent writes the fix">
    Your agent uses those findings to write the change. The fix comes from your agent's model — not from kipmox.
  </Step>

  <Step title="verify_fix — kipmox checks it">
    Before the fix is applied, kipmox re-runs the rules on the original vs proposed code, catches regressions and band-aid fixes, and returns a verdict.
  </Step>
</Steps>

## Two tools

| Tool | What it does |
| - | - |
| [`analyze_code`](/agent/tools#analyze-code) | Deterministic Salesforce static analysis over Apex and LWC. Returns structured findings with per-rule remediation hints. |
| [`verify_fix`](/agent/tools#verify-fix) | Independently verifies a proposed fix **before it's applied** — re-runs the rules, detects regressions and band-aid fixes, and returns a verdict. |

## Why "deterministic" matters

`verify_fix` doesn't ask a model to grade its own work. The checks are deterministic — they run the same way every time, regardless of which model wrote the fix. That's what makes verification a real signal instead of a second opinion.

<Info>
  kipmox Agent also honors **your project's Salesforce Code Analyzer configuration**, including custom PMD rules — and applies it when it *verifies* your agent's fix, not just when it analyzes. See [Custom rules](/agent/tools#custom-rules).
</Info>

## Who it's for

* Salesforce developers using an AI coding agent who want a **deterministic review + verification** step in the loop.
* Teams who want AI-written Apex and LWC checked against **their own ruleset** before it lands.
* Anyone who wants kipmox's guardrails **without the VS Code extension** — any MCP-capable agent works.

## Access

kipmox Agent needs a paid kipmox plan. It's included with **Pro** and **Enterprise**, or available on its own as the **kipmox Agent** tier. See [Plans & Limits](/help/plans-and-limits), or get started at [kipmox.com/agent](https://kipmox.com/agent).

<CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/agent/quickstart">
    Sign in, add the MCP server, and run your first analyze → fix → verify
  </Card>

  <Card title="Tools" icon="screwdriver-wrench" href="/agent/tools">
    analyze\_code, verify\_fix, verdicts, and custom rules
  </Card>

  <Card title="How Fixes Are Verified" icon="circle-check" href="/trust/how-fixes-are-verified">
    The checks behind every verdict
  </Card>

  <Card title="Troubleshooting" icon="life-ring" href="/agent/troubleshooting">
    Sign-in, server setup, and custom-rule issues
  </Card>
</CardGroup>