DEV Community: Emmanuel Ene Rejoice Gideon

I Dogfooded My 49KB AI Coding Agent — It Fought Back 5 Times

Emmanuel Ene Rejoice Gideon — Sat, 05 Sep 2026 16:13:58 +0000

Yesterday I wrote about building an AI coding agent that ships in under 50KB. A fair comment kept coming up: "sounds cool, but does it actually work?"

Fair question. So today I did the thing every builder dreads: I dogfooded the version sitting on npm — not my dev repo, the actual npm i -g stew-ai package a stranger would get.

It fought back. Five times.

Round 1: The auth flow was broken

First command a new user runs after installing is either stew register or stew login <key>. Here's what 2.7.5 actually did:

$ stew whoami
❌ Error: apiKey is not defined

A ReferenceError. In production. The whoami command referenced an apiKey variable that was scoped to a different function — a classic closure bug that my offline tests never caught, because the broken path only triggers when a real key exists in ~/.stew/config.json.

While fixing it, I found four more:

whoami crash — the scope bug above. apiKey lived in authCommand, not whoamiCommand.
stew login <key> ignored the key — the CLI treated the first argument as the action name, so your key was parsed as a command and you got dropped into the interactive prompt anyway.
register() sent the wrong field name — the SDK posted full_name while the FastAPI backend expects name. A silent 422.
usage() authenticated wrong — the SDK sent the key as a Bearer header while the endpoint expects an api_key query param. whoami could never show your plan.
422 errors printed as [object Object] — FastAPI validation errors were stringified raw, so users saw garbage instead of "Missing/invalid field: name".

Five bugs. All in the first 60 seconds of a new user's experience. All invisible in my test suite.

This is why you dogfood the published artifact. My tests ran against source files; the bug shipped anyway.

The fix: same day, with receipts

All five fixes went into v2.7.6, plus five regression tests so each bug can never quietly return:

check('auth: whoamiCommand no longer throws ReferenceError (apiKey scoped)', ...);
check('auth: direct alias routing — "stew login <key>" passes key, not action', ...);
check('SDK: register() sends "name" field (backend contract), not full_name', ...);
check('SDK: usage() passes api_key as query param (backend contract)', ...);
check('StewError: FastAPI 422 detail array becomes readable message', ...);

26/26 pass, package is 49.1KB — still under the 50KB budget, still zero dependencies.

And the flow a user actually experiences now:

$ stew login stew_O9X...
✅ Logged in! API key saved to ~/.stew/config.json

$ stew whoami
👤 S.T.E.W Account
  API Key: stew_O9X...P8b2
  Plan: free

Round 2: Actually building something

Auth works — fine. But the real question was whether the agent could code. So I opened the REPL (stew code) and asked it to build a single-file landing page for a fictional Afrobeat festival in my hometown:

Build a stunning single-file index.html for a fictional Afrobeat concert in Enugu, Nigeria called 'Enugu Sound Fest 2026'. Dark theme, orange/green gradient accents, animated hero, artist lineup grid, live countdown timer, tickets section. Pure HTML/CSS/JS, no external dependencies.

It generated the whole page — hero, lineup grid, countdown, tickets. But when I opened the file, one line was broken:

el.textContent = ${d}d ${h}h ${m}m ${s}s;   // ← template literal lost its backticks

So I did what any user would do: went back into the REPL and told it what was wrong. It rewrote the line with plain concatenation and applied the patch:

And the rendered result, screenshotted from the actual file it wrote:

Build → spot the bug → describe it → agent fixes it → verify. That loop worked, end to end, over the free API tier.

Honest friction notes

Dogfooding also surfaced things I'm not proud of, so you get those too:

The file-apply format is too picky. The CLI only writes files when the model formats code blocks in one exact way (a // filepath: marker on the first line). The model often guessed a different format and the change silently didn't apply. It worked when I told it the exact format — but a user shouldn't have to know the magic words. That's next on the fix list.
A queued message wedged the REPL while a response was streaming — I had to restart the session. Input needs a lock during streaming.

Both of those are now tracked with the same rigor as the auth bugs: found by using the thing, not by staring at the code.

The takeaway

If you build a tool for other people, the most valuable hour you can spend is being one of them — installing the published package, running the first three commands a stranger would run, and believing every error message you see.

The five worst bugs in my auth flow survived 21 passing tests and two published versions. They died in one afternoon of pretending to be my own user.

Package: https://www.npmjs.com/package/stew-ai (v2.7.6, 49.1KB, zero deps)
Source: https://github.com/emmanuelerogian723-alt/stew-ai
Free key: https://stew-agent.onrender.com

Go break it. I'll fix what you find.

I Built an AI Coding Agent That Ships in Under 50KB — Zero Dependencies, Runs on a Phone

Emmanuel Ene Rejoice Gideon — Sat, 05 Sep 2026 15:13:50 +0000

I'm a developer from Enugu, Nigeria. Most devs I know here code on second-hand laptops — and a lot of them code straight from cheap Android phones using Termux.

So when I set out to build an AI coding agent for the African market, "first, install 400 npm packages" was never going to work. Neither was "you need a $2,000 laptop."

I set myself one hard rule: the entire package must stay under 50KB, with zero dependencies.

Meet Stew Code — a terminal AI agent that hits that rule and still does all of this:

👀 Vision — /image screenshot.png and it actually sees the screenshot
🌐 Headless browsing — cookies, forms, clicks, login flows. No Puppeteer
📸 Screenshots & screen recording — web pages or your whole screen
🔒 Security scanning — audits your codebase for real issues
🐝 Swarm mode — a planner splits your task, parallel agents execute it
🗣️ Voice — it reads responses out loud using OS text-to-speech
🔌 MCP client — connect any Model Context Protocol server
🛠️ Self-healing installs — missing Chromium? It auto-installs it, and warns you first: "~250MB download, battery at 12% — plug in before we start"

Quick start

npm i -g stew-ai

That's the whole install. 49KB. On a bad 3G connection it takes seconds, not minutes.

Grab a free API key at https://stew-agent.onrender.com (free tier: 1,500 calls/month), then:

stew auth login   # paste your key
stew code         # launch the REPL

And you're in.

What the brain looks like

Honest disclosure: the intelligence runs on my Stew Agent API — the CLI is the lightweight terminal client. The API is OpenAI-compatible (/v1/chat/completions), so you can also point any OpenAI-shaped tool at it by changing the base URL.

The API side packs 59 skills, a 100-agent swarm, 12 personas, and 6 AI providers (Groq, Mistral, NVIDIA NIM, OpenRouter, HuggingFace, OpenAI) with automatic failover — if one provider goes down, it switches mid-flight. And when you npm i -g stew-ai, none of that bloat lands on your machine. One package. Zero dependencies.

The zero-dependency receipts

How do you fit vision, browsing, screenshots, recording, a REPL, and an MCP client into 49KB with zero deps?

Node 18+ native fetch — no axios, no node-fetch
Raw ANSI escape codes for all terminal coloring
Hand-written argument parsing
Chrome headless --screenshot for page shots; OS-native tools (scrot, screencapture, ffmpeg, PowerShell) for everything else
The npm package ships minified; the GitHub repo stays fully readable
A regression suite runs offline against mocks, so nothing breaks silently between releases

Try it on your phone (yes, really)

# Install Termux from F-Droid, then:
pkg install nodejs
npm i -g stew-ai
stew code

That's an AI coding agent with vision and web access running on an Android phone. That's the entire point — AI tooling shouldn't require hardware most of the world can't afford.

Stew Code 😋 Agent on terminal

Emmanuel Ene Rejoice Gideon — Thu, 03 Sep 2026 09:05:35 +0000

🚀 49.9KB. Zero dependencies. Built in Nigeria 🇳🇬
Stew Code is the free AI coding agent rivaling Copilot & Cursor — without the $20/mo price tag.
Terminal-native. Built-in security scanner. Self-verifying code. Marathon mode. 6 AI providers with auto-failover. 16+ self-forging skills.
One command:
bash
npm install -g stew-ai

🔗 Blog: https://erogian.vercel.app/blog/stew-code-free-ai-coding-agent-nigeria-2026
🔗 Site: https://emmanuelerogian723-alt.github.io/stew-ai/
📦 npm: pkg:npm/stew-ai@2.5.0
Built in Africa. For the world. 🌍

Stew Code 😋 Agent on terminal

Emmanuel Ene Rejoice Gideon — Thu, 03 Sep 2026 08:57:02 +0000

DEV Community: Emmanuel Ene Rejoice Gideon

I Dogfooded My 49KB AI Coding Agent — It Fought Back 5 Times

Round 1: The auth flow was broken

The fix: same day, with receipts

Round 2: Actually building something

Honest friction notes

The takeaway

I Built an AI Coding Agent That Ships in Under 50KB — Zero Dependencies, Runs on a Phone

Quick start

What the brain looks like

The zero-dependency receipts

Try it on your phone (yes, really)

Links

Stew Code 😋 Agent on terminal

Stew Code 😋 Agent on terminal