# Dezycro

> Coding Agent Ops — the operations layer for AI coding agents. Dezycro provisions agents with the context they need, lets you direct them while they run, verifies every change against your spec, and keeps an auditable workbook of every decision. It catches intent violations — a shipped behavior that doesn't match what was asked — before they reach your users.

An intent violation is any change where what the agent shipped drifts from what was actually asked: a missing endpoint, a skipped step, a quietly dropped requirement. Dezycro sits around the coding agent your team already uses, gives agents the context to build from how your product actually works, lets you redirect them mid-flight, and verifies that "done" actually matches the spec — surfacing intent violations before they reach users.

## Your code stays yours

Dezycro never accesses your source code. It connects through the Model Context Protocol (MCP), and verification runs against your running API — not your repository. Your specs, journeys, and workbook live in Dezycro; your source code, repository, and secrets stay in your environment and never leave it.

## How it works

- **Provision** — Journeys, real product behavior, your specs, and notebooks become context your agents read before they touch code, so they build from how your product actually works, not from guesses.
- **Direct** — A per-feature command inbox lets you task and redirect agents mid-flight: hand off new instructions, invalidate stale work, and keep every feature's agents pointed at the right thing.
- **Verify** — Auto-generated coverage checks every requirement your spec promises and catches intent violations — false completions, silent regressions, and drift from intent — surfaced through the Blocked False Completion Rate metric.
- **Observe** — Every decision, assumption, task, and blocker an agent records lands in one workbook, tied to the commit and the verification that proves it. Your team and your agents read from the same source of truth.

## Integrations

- **Claude Code** — Dezycro connects to your coding agent over MCP. More integrations coming.

## Pricing

Billing is monthly. The public pricing page is at https://app.dezycro.ai/pricing.

- **Free — $0/mo** — 1 user, 1 workspace, 1 project, 3 features, 2M tokens/mo.
- **Pro — $20/seat/mo** — unlimited projects and features, 5 workspaces, 15M tokens/seat/mo.
- **Team — $39/seat/mo (3-seat minimum)** — unlimited workspaces, 40M tokens/seat/mo, SSO, RBAC, audit log, and unlimited free Viewers.
- **Enterprise — custom** — bring-your-own-key (BYOK), SAML/SCIM, and SLA.

## Documentation

Full product documentation lives at https://docs.dezycro.ai. For agents that want to read more:

- [Machine-readable documentation index](https://docs.dezycro.ai/llms.txt): project description plus every docs page grouped by section, each linking to its markdown.
- [Full documentation (single file)](https://docs.dezycro.ai/llms-full.txt): the entire documentation site as a single markdown document for a one-shot read.

## Pages

- [Landing](https://app.dezycro.ai/): hero, the security model (your code stays yours), the problem, how it works (Provision / Direct / Verify / Observe), the Blocked False Completion Rate metric, pricing, and FAQ.
- [Pricing](https://app.dezycro.ai/pricing): full plan comparison — Free, Pro, Team, Enterprise (monthly billing).
- [Get started](https://app.dezycro.ai/register): Create a workspace — free tier, no credit card.
- [Login](https://app.dezycro.ai/login): Sign in to an existing workspace.
- [Terms](https://app.dezycro.ai/terms): Terms of service.
- [Privacy](https://app.dezycro.ai/privacy): Privacy policy.