Gemini Executive Synthesis

Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script.

Technical Positioning

A highly portable, dependency-free coding agent harness built with system primitives (sh, curl, awk), offering Anthropic/OpenAI integration and a suite of 7 tools for interactive and automated coding tasks.

SaaS Insight & Market Implications

Pu.sh demonstrates a radical approach to AI agent development, prioritizing extreme portability and minimal dependencies by leveraging core shell utilities. This project highlights a counter-trend to complex, framework-heavy AI solutions, proving that powerful agent capabilities can be achieved with a lean footprint. The ability to integrate with major AI models (Anthropic, OpenAI) and provide essential tools (bash, read, write, edit) within 400 lines of shell script is a significant technical feat. This addresses a developer pain point of bloated environments and complex setups, offering a "build your own harness" philosophy. The market implication is the potential for highly embeddable, resource-efficient AI agents, suitable for constrained environments or as foundational components in larger systems where overhead must be minimized.

Proprietary Technical Taxonomy

Raw Developer Origin & Technical Request

Hacker News May 1, 2026

Show HN: Pu.sh – a full coding-agent harness in 400 lines of shell

I originally was just messing with pi-autoresearch. Gave it a sample task to build the most portable coding agent.First cut was 6 KB of shell. Great for one-shots, unusable interactively. I was shocked it actually worked.Started building up -- adding features — but with a self-imposed rule: no new dependencies, and sub 500 LOC. This thing had to be truly portable. Just sh, curl, awk. System primitives only.Which means I did some genuinely disgusting things in awk, including JSON parsing and the OpenAI
Responses tool loop with reasoning items carried across turns.It's now ~400 lines. In the box: Anthropic + OpenAI, 7 tools (bash, read, write, edit, grep, find, ls),
REPL, auto-compaction, checkpoint/resume, pipe mode, 90 no-API tests. Not in the box: TUI, streaming,
images, OAuth, Windows, dignity.Two honest things:1. I stole/modified the system prompt and the architecture. Pi/Claude/Codex wrote the awk. I cannot read most of this code. This wasn't possible for me a year ago.2. Heavily inspired by Pi (pi.dev) — same 7-tool surface, same exact-text edit model. Credit where it's
due. Pi is awesome -- you should probably use them.The agent loop itself is tiny. Almost everything else in a "real" agent CLI is DX and hardening. You can
probably build your own harness exactly how you like it. Mario Zechner's AI Engineer talk on taking back control of
your tools nudged me here.The name is because it's a .sh file. The other thing it sounds like is, regrettably, also accurate.

View Raw Source

Developer Debate & Comments

Amber-chen • May 1, 2026

I like the small-surface-area approach. The question I’d use to evaluate this is how well the harness records/replays tool calls and failure modes, since that is where debugging agent behavior usually gets messy.

zackham • May 1, 2026

if you are looking for a customizable agent harness that provides everything except a tui (cli, acp, embeddable python package) i built and use this for a handful of projects where i needed a vendor-agnostic replacement for claude agent sdk: https://github.com/zackham/aloop

0gs • Apr 30, 2026

why the LOC limit? just because?

mjuarez • Apr 30, 2026

Could you post the initial 6KB version? I'd love to have a version that I can actually read and understand. I don't care if it's 2K LOC.

cachius • Apr 30, 2026

Does it work in just-bash?https://github.com/vercel-labs/just-bash/blob/main/packages/...

tecoholic • Apr 30, 2026

Kudos. I set on this exact journey a couple of days back and Pi is what I started reading for inspiration as well. I really can't stand the text boxes and the animations of the mainstream harnesses.

kkovacs • Apr 30, 2026

I love this. This is the grown-up brother of my one-liner bash+python athttps://github.com/kkovacs/kkrc/blob/ecff1f65d5ccfa24d5f6695...:)

flog • Apr 30, 2026

I'd love to learn how to build something like this, but unfortunately the code is unreadable.If author could write a heavily commented version that would be amazing.

ricardobeat • Apr 30, 2026

Really like the looks of it, but minifying the code to achieve the “400 lines” marketing gimmick is a huge turn-off.It’s also a security nightmare, and ensures it remains 100% vibe coded. Would rather have a readable source with an honest line count.

petcat • Apr 30, 2026

your 500 LOC requirement has made this thing completely and unnecessarily inscrutable and unreadable.Pass.

Frequently Asked Questions

Market intelligence mapped to Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script..

What problem does Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script. solve?

Based on our AI analysis of the original developer request, its primary technical positioning is: A highly portable, dependency-free coding agent harness built with system primitives (sh, curl, awk), offering Anthropic/OpenAI integration and a suite of 7 tools for interactive and automated coding tasks.

How is the developer community reacting to Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script.?

Yes, we have tracked 21 direct responses and active debates regarding this specific topic originating from Hacker News.

What architecture is tied to Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script.?

Our proprietary extraction maps Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script. to adjacent architectural concepts including coding-agent harness, shell script, portable, no new dependencies.

How does the GitHub community build with Pu.sh, a full coding-agent harness implemented in ~400 lines of shell script.?

Yes, open-source adoption is correlated. An active project titled 'wangziqi06/724-office' explores similar frameworks: 7/24 Office — Self-evolving AI Agent system. 26 tools, 3500 lines pure Python, MCP/Skill plugins, three-layer memory, self-repair, 24/7 production.

Engagement Signals

Upvotes

Comments

Cross-Market Term Frequency

Quantifies the cross-market adoption of foundational terms like sh and Anthropic by tracking occurrence frequency across active SaaS architectures and enterprise developer debates.