simplicity over complexity

A lightweight, local-first coding agent.

Built by people who like quiet tools. Bring your own key, run it locally, or grab a prepaid HamrPass. No subscription, no login.

github ↗

~/projects/api · codehamr

What is codehamr?

A reliable hammer for your code.
Plain-text config, no rabbit holes.
Local-first, OpenAI-compatible, open standards only.

What it isn't.

No monthly subscription.
No plugin sprawl, no extension zoo.
No telemetry, no proprietary endpoints.

we love local

Ollama. vLLM. llama.cpp.

We speak the standard API they all share. Nothing more, nothing less.

pick the right tool

When to pick what.

frontierClaude Code and Codex are great first-party agents, but they're expensive subscriptions with session timeouts.

opencode↗Great Swiss army knife of coding, if you want to embrace the complexity.

pi coding agent↗Lighter than opencode, if you like customization with extensions, skills, and themes.

codehamr↗Simplicity over complexity, if you'd rather code than configure.

free, DIY, or curated.

Three ways to run.

free

Local

Built from scratch for local LLM coding. A ~30B class model like qwen3.6:27b codes smoothly on a beefy Macbook or 32 GB+ VRAM. No GPU? That's what HamrPass is for.

DIY

Bring your own key

Any OpenAI-compatible endpoint works. OpenRouter, your own proxy, your models, your bill. Flexible where it counts, but we avoid proprietary endpoints that lock you in.

$20 prepaid pass curated

HamrPass

Prepaid. No login, no auto-renew, no subscription. Burn tokens when you need them. We did the model hunt for you. Optimized for codehamr, ready out of the box.

Watch for launch ↗

setup

Setup, in plain text.

install

curl -fsSL https://raw.githubusercontent.com/codehamr/codehamr/main/install.sh | bash

Generated at .codehamr/config.yaml on first start, in your working directory. Yours to edit.

.codehamr/config.yaml

active: local
models:
    local:
        llm: qwen3.6:27b
        url: http://host.docker.internal:11434
        key: ""
        context_size: 64000
    openai:
        llm: gpt-5.5
        url: https://api.openai.com
        key: sk-...
        context_size: 128000
    hamrpass:
        llm: hamrpass
        url: https://codehamr.plaxtoris.de
        key: hp_...
        context_size: 512000

Project rules or custom system prompts? They go into .codehamr/CODEHAMR.md. Optional, additive, one file. The single point of adjustment.

install

It's hamrtime.

Get the binary. Run it locally. Pick up a HamrPass when local can't keep up.

github ↗