---
title: DIY harness
slug: diy-harness
url: "https://toolweight.com/options/diy-harness"
homepage: "https://toolweight.com/compare/coding-agents"
categories: coding-agents
last_verified: 2026-07-23
license: CC-BY-4.0
---

# DIY harness

> Your own loop over a model API with an agent SDK, the build-it-yourself reference point.

The baseline: a few hundred lines over whichever agent SDK your provider ships, every vendor on this page has one, and this row deliberately endorses none of them, with the tools your codebase actually needs and none of the ones it does not. This is genuinely the right answer for narrow, repeated tasks, a migration bot, a test-backfill job, where a general harness spends most of its context rediscovering your conventions. It is the wrong answer for open-ended engineering, because you will spend a quarter rebuilding permissions, checkpointing and context management badly.

## Identity

|  |  |
| --- | --- |
| Name | DIY harness |
| Company | DIY harness |
| One-liner | Your own loop over a model API with an agent SDK, the build-it-yourself reference point. |
| Site | https://toolweight.com/compare/coding-agents |
| Open source | Yes |
| Compared in | 1 |

## Where it is compared

### [Coding agents](https://toolweight.com/compare/coding-agents)
Ranked **#12 of 17** on default weights.
| Field | Value | Confidence | Verified | Source | Note |
| --- | --- | --- | --- | --- | --- |
| Surface | CLI | Inferred | - | - | Whatever you build it as. |
| Default model | Whichever you wire up | Inferred | - | - | - |
| Model breadth | any provider / router | Inferred | - | - | - |
| Repo indexing | manual context only | Inferred | - | - | Until you build something better, which is most of the work. |
| Plan mode | ○ | Inferred | - | - | - |
| Checkpoint & undo | ○ | Inferred | - | - | - |
| Headless / CI | ● | Inferred | - | - | It is a script; headless is the default. |
| Local models | ● | Inferred | - | - | - |
| Where it stops | - | Unknown | - | - | There is no ceiling, it goes exactly as far as you build, and you own every rung. Not scored, because ranking your own code against shipped products would be meaningless. |
| MCP | ◐ | Inferred | - | - | Most agent SDKs ship an MCP client. |
| Subagents | ◐ | Inferred | - | - | - |
| Hooks | ● | Inferred | - | - | It is your code; every line is a hook. |
| Skills / commands | ○ | Inferred | - | - | - |
| Background agents | ◐ | Inferred | - | - | - |
| Opens PRs | ◐ | Inferred | - | - | - |
| Permissions | applies everything | Inferred | - | - | Unless you build a permission layer, which almost nobody does properly. |
| BYOK | ● | Inferred | - | - | - |
| Licence | - | Unknown | - | - | Yours. |
| Pricing model | BYOK tokens only | Inferred | - | - | - |
| Entry price | $0 /seat/mo | Inferred | - | - | Zero licence cost, non-zero engineering cost, budget a fortnight before it is useful and a quarter before it is safe. |
| Token markup | ○ | Inferred | - | - | - |
| GitHub stars | - | Unknown | - | - | - |
| Terminal-Bench | - | Unknown | - | - | - |
| Bench model | - | Unknown | - | - | - |
| Positioning | The reference point: what you get for a few hundred lines and an API key. | Inferred | - | - | - |
| First release | - | Unknown | - | - | - |

**Verdict.** Right for narrow, repeated jobs, a codemod runner, a test-backfill bot, where a general harness wastes most of its context rediscovering conventions you could have hardcoded. Wrong for open-ended engineering, where you will spend a quarter rebuilding permissions, checkpointing and context management worse than the incumbents.

## Alternatives

- [Aider](https://toolweight.com/options/aider), Minimal terminal pair-programmer with a tree-sitter repo map and automatic git commits.
- [Amp](https://toolweight.com/options/amp), Sourcegraph's opinionated agent with no model picker and an ad-supported free tier.
- [Claude Code](https://toolweight.com/options/claude-code), Anthropic's terminal-first agent, with IDE, web and GitHub Action surfaces.
- [Cline](https://toolweight.com/options/cline), Open-source VS Code and JetBrains agent built around an explicit plan/act split.
- [Codex CLI](https://toolweight.com/options/codex-cli), OpenAI's open-source Rust agent, with the strongest local sandboxing on this page.
- [Cursor](https://toolweight.com/options/cursor), VS Code fork with a real codebase index, background cloud agents and a CLI.
- [Devin](https://toolweight.com/options/devin), Fully delegated cloud engineer: file a task in Slack or the web app, get a PR back.
- [Gemini CLI](https://toolweight.com/options/gemini-cli), Apache-2.0 terminal agent from Google with an unusually generous free tier.
- [GitHub Copilot (Agent HQ)](https://toolweight.com/options/github-copilot), Copilot's coding agent plus a control plane for running third-party agents on your repos.
- [Goose](https://toolweight.com/options/goose), Block's Apache-2.0 agent, MCP-native from the start, as both a CLI and a desktop app.
- [Grok Build](https://toolweight.com/options/grok-build), xAI's coding harness, launched July 2026, on this page but not yet verified.
- [Kilo Code](https://toolweight.com/options/kilo-code), Open-source VS Code agent in the Cline/Roo lineage, with an orchestrator mode and a CLI.

## Licence and attribution

Data from toolweight (https://toolweight.com), licensed CC-BY-4.0.

- Licence: [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)
- Canonical HTML: https://toolweight.com/options/diy-harness
- Machine-readable: https://toolweight.com/options/diy-harness.md · https://toolweight.com/api/v1 · https://toolweight.com/mcp
- toolweight takes no affiliate revenue and sells no placements. Corrections: https://toolweight.com/suggest