# Hanzo Kata — Autonomous coding agent & CI/CD loop

> Autonomous test repair, CI/CD triage, and repository maintenance. Powered by Kai typed decisions and Enso model routing.

Hanzo Kata · Autonomous Coding Agent

# Test. Repair. Verify.Autonomous loop.

Kata closes the loop on software maintenance: triaging failing CI/CD runs, locating regressions, executing surgical code repairs, and validating green test suites under strict deterministic governance.

[Get API Key](https://platform.hanzo.ai/keys)[Explore GitHub Examples →](https://github.com/hanzoai/kata)

[Read the Docs →](https://docs.hanzo.ai/docs/coding)

terminal

$kata run --suite tests/unit

12 passed · 1 failed (test_token_expiry)

$kata triage

Kai decision: patch · confidence: 0.96 · root cause: off-by-one jitter

$kata repair --model enso-ultra

synthesized AST edit in auth/tokens.py · re-running test suite...

13 passed · 0 failed · 100% green verification

$

The Closed-Loop Cycle

## From failing trace to verified PR.

Kata never submits speculative code. Every fix is triaged by Kai, synthesized via Enso, and verified against local or cloud test suites.

01. Observe Failure

### Detect & Capture

Extract stack traces, exit codes, and test logs from pytest, cargo test, vitest, or go test.

02. Kai Decision Head

### Deterministic Triage

Kai scores whether to retry, patch code, roll back, or escalate to human review with zero output tokens.

03. Enso Model Routing

### Optimal Synthesis

Enso routes code edit generation to the best frontier or open-weight model based on quality, latency, and budget.

04. Closed-Loop Verify

### Green Suite Guarantee

Executes candidate diffs in an isolated microVM sandbox. Only PRs with 100% green verification land.

Capabilities

## Deterministic engineering. Zero hallucinations.

Engineered for high-reliability CI/CD pipelines, autonomous bug repair, and multi-agent systems.

### Autonomous Test Repair Loop

Identifies regressions and flaky tests, locates the root cause in your AST, applies surgical diffs, and verifies green passes.

### Deterministic Triage via Kai

Stop parsing conversational LLM prose. Kai returns typed probabilities over discrete actions (patch, rollback, retry, approve).

### Intelligent Model Routing via Enso

Balances cost and accuracy on every iteration using enso-auto, enso-flash, or enso-ultra presets.

### CI/CD Pipeline Integration

Drop into GitHub Actions, GitLab CI, or terminal pre-push hooks with zero configuration overhead.

Kata · Autonomous Coding Agent

## Autonomous test repair specifications, endpoints, and multi-language SDKs.

Closed-loop agent runtime designed for high-consequence CI/CD environments. Combines Kai typed decision heads with Enso multi-tier model routing and sandbox execution.

Model Card Specifications

kata

Wire EndpointCLI & Agentic Loop · api.hanzo.ai

Autonomous Cycle

Capture → Kai Triage → Enso Synthesis → Sandbox Verify

Decision Engine

Kai typed decision model (determines next action)

Model Routing

Enso microsecond router (enso-auto, enso-flash, enso-ultra)

Test Runners

pytest · vitest · cargo test · go test · custom TAP

Sandbox Isolation

Isolated Linux container / macOS seatbelt

Multi-Language SDKs

Python · TypeScript · Go · Rust · cURL

License & Source

Apache-2.0 Open Source on GitHub

[Get API Key](https://platform.hanzo.ai/keys)[🤗 hanzoai on HF Hub](https://huggingface.co/hanzoai)[HF Space](https://huggingface.co/spaces/hanzoai/guard-arena)[Documentation](https://docs.hanzo.ai/docs/coding)

Multi-Language SDKs & OSS Examples

[🤗 Hugging Face](https://huggingface.co/hanzoai)[GitHub Repo](https://github.com/hanzoai/kata)

Runnable examples across all tier-one runtimes. Zero-install authentication with hanzo auth login or environment variable HANZO_API_KEY.[Python Loopkata_loop.py](https://github.com/hanzoai/kata/tree/main/examples/python)[TypeScript Loopkata_loop.ts](https://github.com/hanzoai/kata/tree/main/examples/typescript)[Go Clientmain.go](https://github.com/hanzoai/kata/tree/main/examples/go)[Rust Clientsrc/main.rs](https://github.com/hanzoai/kata/tree/main/examples/rust)[cURL Scriptcurl.sh](https://github.com/hanzoai/kata/tree/main/examples)

[Modular Rust Recipes Crate (kata-recipes)Modular production blueprints, enterprise crates, and automated verification suites.](https://github.com/hanzoai/kata/tree/main/recipes)python

```
import os from openai import OpenAI client = OpenAI( base_url="https://api.hanzo.ai/v1", api_key=os.environ["HANZO_API_KEY"], ) # Kata autonomous triage loop over failed test response = client.chat.completions.create( model="enso-ultra", messages=[ {"role": "system", "content": "You are Kata, the autonomous test repair agent."}, {"role": "user", "content": "Triage failed test: KeyError 'user_id' in auth/tokens.py:42"} ], ) print(response.choices[0].message.content)
```

Enterprise Solutions

## Battle-tested autonomous repair blueprints for enterprise software systems.

Eliminate flaky tests, accelerate legacy refactoring, and reduce CI/CD test failure triage times from hours to seconds.

04AGENTIC CODING

Stack: Kata + Gen + Enso

### Autonomous CI/CD Test Repair & Regression Triage

Hooks into GitHub Actions and GitLab CI. When a unit or integration test fails, Kata captures the trace, triages root cause via Kai, synthesizes an AST patch via Enso, and verifies green passes.Value Impact:$55M–$552M capacity value

Grounding Documentation & Code[HIP-0510 (Closed-Loop CI/CD)](https://hips.hanzo.ai/docs/hip-0510-enso-learned-router)[Closed-Loop Autonomous Test Repair](https://hanzo.blog/blog/2026-10-16-closed-loop-agentic-repair/)[recipes/src/repair.rs](https://github.com/hanzoai/kata/tree/main/recipes)

[Explore Repair Recipe](https://github.com/hanzoai/kata/tree/main/recipes)

05MODERNIZATION

Stack: Gen + Kata

### Legacy Modernization & Service Regression Suites

Re-platforms monolithic backends into modern Go/Rust microservices. Kata autonomously generates comprehensive regression test suites to prove exact functional parity.Value Impact:10x faster service migration

Grounding Documentation & Code[HIP-1334 (Unified Test Runtime)](https://hips.hanzo.ai/docs/hip-1334-research-unified-runtime)[Enterprise Diligence & AI Solutions](https://hanzo.blog/blog/2026-09-21-ai-pain-points-investment-firms/)[recipes/src/modernize.rs](https://github.com/hanzoai/kata/tree/main/recipes)

[Explore Modernization Recipe](https://github.com/hanzoai/kata/tree/main/recipes)

15DETERMINISTIC QA

Stack: Kai + Kata

### Flaky Test Elimination & Sandbox Replay

Isolates non-deterministic test flakes in sandboxed containers. Triages root causes (race condition, timeout, clock skew) via Kai typed heads and synthesizes deterministic assertions.Value Impact:100% green build guarantee

Grounding Documentation & Code[HIP-1145 (Deterministic Replay)](https://hips.hanzo.ai/docs/hip-1145-research-experiment-record)[Autonomous Agents in CI/CD](https://docs.hanzo.ai/docs/coding)[recipes/src/replay.rs](https://github.com/hanzoai/kata/tree/main/recipes)

[Explore Sandbox Recipe](https://github.com/hanzoai/kata/tree/main/recipes)

Research & Architecture

## Closed-loop verification, deterministic evaluation, and agentic DevOps.

Read engineering papers, verification benchmarks, and architecture write-ups from the Kata systems team.

[Architecture2026-10-16 ### Closed-Loop Autonomous Test Repair: From Stack Trace to Green PR Why generative completions fail in CI/CD without deterministic evaluation heads. How Kata combines Kai decisions, Enso model routing, and microVM sandboxes.Read Article](https://hanzo.blog/blog/2026-10-16-closed-loop-agentic-repair/)

[Enterprise AI2026-09-21 ### Top Pain Points for Investment Firms That Hanzo AI Solves Identify key operational challenges in engineering and diligence — and discover how Hanzo’s sovereign AI cloud and agentic workflows eliminate maintenance toil.Read Article](https://hanzo.blog/blog/2026-09-21-ai-pain-points-investment-firms/)

[Engineering Guide2026-08-30 ### Autonomous Coding Agents in High-Consequence CI/CD Pipelines Best practices for deploying autonomous test repair loops inside GitHub Actions, GitLab CI, and corporate Jenkins clusters under strict deterministic governance.Read Article](https://docs.hanzo.ai/docs/coding)

[Read more on Hanzo Blog →](https://hanzo.blog)

Hanzo Intelligence Ecosystem

## Complementary engines. Built to compose with Kata.

Kata focuses on continuous testing, repair, and green verification; pair it with autonomous feature engineering, universal harnesses, and decision heads.

Hanzo Dev

CODING AGENT

Autonomous software engineer that designs, implements, and tests features from plain English.How they pair: Dev handles forward feature synthesis; Kata continuously maintains the test suites and fixes regressions.

Key Benchmark / MetricAutonomous Pull Requests

[Explore Hanzo Dev](https://hanzo.ai/dev)

Hanzo Gen

AGENT HARNESS

Universal CLI harness for running any coding agent on Hanzo cloud rails.How they pair: Gen launches Kata inside terminal workflows or integrates Claude Code and Codex with Kata verification.

Key Benchmark / MetricUniversal MCP Tools

[Explore Hanzo Gen](https://hanzo.ai/gen)

Hanzo Kai

DECISION MODEL

Calibrated multimodal decision intelligence with zero output token fees.How they pair: Kata uses Kai as its deterministic decision head to triage failures into retry, patch, or rollback.

Key Benchmark / Metric85 ms CPU Latency

[Explore Hanzo Kai](https://hanzo.ai/kai)

Hanzo Enso

LEARNED ROUTER

Microsecond learned utility router that balances quality, latency, and cost across 100+ frontier models.How they pair: Kata routes repair diff synthesis through Enso to balance cost and code quality on each retry.

Key Benchmark / Metric12 µs Utility Routing

[Explore Hanzo Enso](https://hanzo.ai/enso)

Questions

## Asked plainly.

How does Kata differ from Hanzo Dev and Gen?

Dev is a conversational coding engineer that plans and writes full features. Gen is the universal CLI harness for any coding agent. Kata is the closed-loop autonomous repair and CI/CD triage engine that continuously keeps test suites green.

Which test frameworks does Kata support?

Kata natively supports pytest, vitest, jest, cargo test, go test, and custom test runners emitting standard TAP or JUnit XML output.

Where can I find the open-source examples and recipes?

The open-source repository is at github.com/hanzoai/kata with runnable examples in Python, TypeScript, Go, Rust, and cURL, along with a standalone Rust recipes crate.

How do I authenticate with Hanzo Cloud?

Log in once with `hanzo auth login` or export `HANZO_API_KEY` from platform.hanzo.ai/keys.

Read

## The agent, the loop, the documentation.

Explore the source code, recipes, and documentation.

### Kata

[hanzoai/kataGitHub Repository.](https://github.com/hanzoai/kata)[Multi-Language SDKs](https://github.com/hanzoai/kata/tree/main/examples)[Rust Recipes Crate](https://github.com/hanzoai/kata/tree/main/recipes)

### Agent Systems

[Hanzo DevOur coding agent.](https://hanzo.ai/dev)[Hanzo GenUniversal harness.](https://hanzo.ai/gen)[KaiDecision engine.](https://hanzo.ai/kai)

### Models

[EnsoModel router.](https://hanzo.ai/enso)[ZenOpen weights.](https://hanzo.ai/zen)[Every model](https://hanzo.ai/models)

[Kata is one of three Hanzo model families. Compare them →](https://hanzo.ai/models)

Paste and ship

Paste this into any agent. It reads the skill manifest and calls Kata from there.

Agent promptCopy

```
Read https://hanzo.ai/skill.md and use Hanzo Kata in my project. Start with: hanzo kata run
```

## Continuous test maintenance on autopilot. Keep your test suites 100% green with Kata.

[Try Hanzo](https://hanzo.ai/login)[Explore Repair Recipes](https://github.com/hanzoai/kata/tree/main/recipes)