blank3rs

blank3rs / heso

Public

The agent-native web engine. No Chromium. No Node. One Rust binary. Search, fetch, parse, JS, DOM, cookies, forms, clicks, sessions — all returning JSON.

10
2
85% credibility
Found May 23, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

Heso is a lightweight headless web engine designed specifically for AI agents. It allows an AI to fetch web pages, run JavaScript, click buttons, fill forms, and extract structured data — all without needing a full browser like Chrome. The key differentiator is reproducibility: workflows can be 'stamped' with a complete record of network requests, then replayed byte-identically later without touching the internet. This makes AI web tasks reliable and verifiable. Heso works as a single Rust binary, a Python library, and a Node.js package, with signed receipts for verification using Ed25519 cryptography. It's currently pre-alpha (v0.0.10) but well-documented.

How It Works

1
🔍 You need an AI to browse the web

Your AI agent needs to look up real websites, read pages, and interact with buttons and forms — but the usual browser tools are complicated.

2
📦 You install heso in one line

A simple one-line install brings a tiny 9 MB program onto your computer. No large downloads, no extra software needed.

3
🌐 Your AI opens any website instantly

When your agent says 'heso open https://news.ycombinator.com', it receives structured data: page title, headings, clickable buttons, and form fields — all numbered and ready to use.

4
Your AI can now do two kinds of browsing
📄
Static mode — fast, no JavaScript

Good for news sites, documentation, and pages that don't need scripts to show content.

⚡
JavaScript mode — runs page scripts

Your agent can click buttons, fill forms, wait for content to load, and see what the page looks like after scripts run.

5
🔒 You can stamp and replay any workflow

When your agent does a sequence of actions — open, click, fill, submit — you can 'stamp' it. The stamp captures everything, so the same sequence runs exactly the same way later, even offline.

6
📝 Results come back as clean JSON

Every page visit returns a consistent JSON structure with title, text, links, forms, and a unique hash — so your agent can compare what changed between visits.

✅ Reliable web browsing for your AI

Your agent can search the web, read any page, click and fill forms — and you can trust the results because they can be verified and replayed exactly.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is heso?

Heso is a headless web engine written in Rust that lets you fetch pages, run JavaScript, click elements, fill forms, and extract data -- all returning JSON. Instead of bundling Chromium or Node, it ships as a single 9.6 MB binary that handles the full browser automation stack. The core workflow centers on verbs like `heso open`, `heso read`, `heso click`, and `heso search`, with Python and Node.js wrappers for programmatic use.

Why is it gaining traction?

The standout feature is reproducibility: every run can be "stamped" into a signed plat that embeds all network responses, then byte-identically replayed offline. This matters for agents that need deterministic, auditable behavior rather than live network calls. The signed receipts add cryptographic verification for compliance-heavy workflows. Performance is aggressive -- cold starts around 77ms, batch operations for 8 URLs in roughly 1.1 seconds. The agent-native design means it was built for AI tool-calling from day one, not retrofitted onto an existing browser.

Who should use this?

AI developers building agents that need reliable web interaction will find the most value here. The reproducibility model suits anyone building audit trails or needing offline replay (think compliance, testing, or deterministic scraping). Python and Node.js bindings make integration straightforward for teams already in those ecosystems. Teams needing pixel-perfect rendering or fighting sophisticated bot detection should look elsewhere -- this is for structured data extraction, not visual testing.

Verdict

Heso solves a real problem with a clean design, but the credibility score of 0.8500000238418579% and single-digit star count signal it's early and unproven at scale. Worth exploring for agent workloads where reproducibility beats live network calls, but don't bet production pipelines on it yet.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.