Scaling your solopreneur business with autonomous agents.
Real artifacts, production workflows, and AEO-ready frameworks. Copy our systems to build a 10-person business, alone.
Field Notes Archive
The Day Our Agent's Context Window Filled Up: A Postmortem
A long-running agent daemon degraded into incoherence as its context window saturated. Here is what broke, why monitoring missed it, and the three-contract architecture that fixed it.
Write the Spec Before the Prompt: A Copy-Paste Template for Spec-Driven Agent Work Orders
Prompts are one-shot utterances; specs are contracts. Here is the minimal spec template we hand to coding agents — plus the scaling rules for when one page is enough and when you need four files.
Valid JSON Is Not Valid Output: Wiring a Technical Contract Gate into CI for AI-Generated Changes
How to validate LLM output in CI: a six-layer contract gate — constrained decoding, agent-loop hooks, schema and policy checks, branch protection, exit-code deploy gates, and provenance — that decides what becomes project state.
Scaling Agentic Infrastructure: A Solopreneur's Guide to 2026
How to architect and scale autonomous AI agents without bankrupting your 1-person company on server costs.
Vercel Is Not a Deployment Contract: What Auto-Deploy Silently Breaks, and How to Fix It
Push-to-deploy on Vercel feels like a contract, but it's a default. A field note on three silent breakage classes — runtime mutation, version skew, asymmetric rollback — and the explicit contracts that close them.
The Zero-Cost Vibecoder Stack: Building a Research Agent for Free
How to build a fully automated research and summarizing agent using 100% free APIs: Firecrawl, Groq, and ArXiv.
The Agentic Software Stack for 1-Person Unicorns in 2026
We reveal the actual 2026 uncrewed technology stack and architecture we use to run a 10-person production operation entirely solo.
One Command, Full Research: Building a Local Knowledge Engine with musu-crawl-ai
How a single Go binary replaces your scattered research workflow — fetching YouTube, Arxiv, GitHub, and the open web into a structured, searchable wiki.
Image Design Contract: Visual Assets & Generators
The definitive technical contract for generating, storing, and mapping blog post images. Do not deviate from these concepts.
Use HTML to Review Agent Output, Not to Replace the Contract
HTML is great for reviewing complex AI agent outputs, but dangerous as a system of record. Learn why the final decision must always export back to Markdown or JSON.
Stateless MCP Servers Can Still Leak Shared State: A Security Contract
An HTTP endpoint might be stateless, but your MCP server might not be. Learn how shared object state in AI agent infrastructure can silently leak data across clients.
The Work Disk Contract: Managing Artifacts in AI Coding Agents
AI coding agents generate artifacts like builds, tests, and logs. Discover why defining a strict 'Work Disk Contract' is crucial to prevent operating system failures and lost evidence.
Vercel Is Not a Deployment Contract
A Coolify migration audit exposed two hidden Vercel assumptions: rewrites that only existed in vercel.json and a build script that depended on a Unix shell. Learn how to verify your deployment contract.
How to Stop AI Agents From Losing Their Memory: The Operating Structure
Long LLM prompts are not operating memory. Learn how to build an artifact-driven memory stack with source notes, specs, and searchable indexes for autonomous AI agents.
DESIGN.md: Turning Visual Taste Into a Strict Agent Contract
Visual design for AI agents fails when built on screenshots and vibes. Learn how DESIGN.md turns UI taste into an inspectable, lintable technical contract.
Software 3.0 Is a Verification Problem
The useful Software 3.0 lesson is not that LLMs replace engineering. It is that faster generation moves the bottleneck to context, review, and evidence.
About Vibecode Town: A Field Log for AI-Assisted Software Engineering
Vibecode Town is an evidence-backed field log tracking AI-assisted software development, agentic workflows, and technical contracts. We focus on verifiable proof, not AI hype.
Frustration Is a Signal, Not the Specification: Engineering Agentic Boundaries
Don't just vent at your AI agent when it makes a mistake. Turn repeated frustration into a strict technical contract, a verifier, and a durable test.
What Vibe Coding Actually Is: Exploration vs Production
Vibe coding is useful for exploration, but production work starts when intent becomes a technical contract the agent can verify against.
Subscribe to the Signal
Get one field note when there is a real artifact worth studying. No hype. No spam.