Ever wondered why your codebase feels like a haunted house — full of ghostly features that worked once, then vanished into debug hell?
That’s vibe coding in action. This wild west of AI-assisted development, where Cursor or Copilot spits out pristine PRs faster than a caffeinated intern, is revolutionizing software speed. But here’s the kicker: it’s birthing graveyards of abandoned production codebases. I’ve seen it. You’ve probably felt it. And now, a 25-year shipping vet has dropped a manifesto to fix it all.
Picture this: teams slashing auth features from weeks to days, applause at sprint reviews, only for some rogue AI tweak to shatter it silently later. No memory. No context. Just breakage.
What the Hell is Vibe Coding, Anyway?
Vibe coding — that’s the term for this stateless sprint from prompt to production. No framework. Pure velocity. The original post nails it with three gut-punch stories. Take this one:
A team uses Cursor to build a new authentication feature in 3 days instead of 3 weeks. The demo is flawless. Sprint review: applause. Two sprints later, a completely unrelated AI-generated change silently breaks the auth flow. Nobody knows why.
Chilling, right? It’s like handing a Ferrari to a kid who can’t read a map — thrilling ride, inevitable crash.
And it’s not isolated. 400-line AI PRs slipping through reviews because they lint clean, only to sprout security holes in prod. Or AI agents flip-flopping on architecture decisions across sprints, leaving schizophrenic code no human touched.
We’ve got Agile for squishy human teams. Microservices playbooks for the cloud shuffle. But for AI? Crickets. Until now.
The Agile Vibe Coding Manifesto: AI’s New Constitution
Enter the Agile Vibe Coding (AVC) Manifesto, an open-source love letter to the original Agile values, turbocharged for AI-human hybrids. Written by a battle-hardened dev who’s danced through waterfall winters and microservices storms, it’s live at agilevibecoding.org. Fork it on GitHub. Contribute. It’s begging for your war stories.
Four core values anchor it:
Accountability over anonymous generation. Traceable intent over opaque magic. Discoverable domain structure over scattered spaghetti. Human-readable docs over ‘the AI knew what it was doing’ vibes.
Then 15 principles that hit like a caffeine IV drip: humans own deployed systems, every change traces to requirements, architecture cages the AI beast, context gets versioned like code.
It’s not anti-AI. Far from it. The author uses these tools daily — calls them transformative. But speed without sustainability? Waste. Pure waste.
My unique take? This echoes the NoSQL boom of the 2010s. Remember? Hype exploded for schemaless bliss, ditching relational rigor. Velocity soared. Then came the maintenance apocalypse — data swamps nobody could query, migrations that ate quarters. Vibe coding is NoSQL 2.0 for code itself. Without AVC, we’ll drown in AI-generated bloat by 2026. But nail this manifesto? AI becomes the ultimate junior dev: tireless, but always leashed.
Why Does Vibe Coding’s Graveyard Loom So Large?
Blame the tools’ amnesia. AI’s stateless — fresh chat, fresh brain dump. No ‘why’ encoded. Humans skim reviews, dazzled by clean diffs. Juniors approve patterns they’ve never touched. Prod implodes.
But wait — isn’t velocity the point? Sure, if your app’s a fireworks show. For business software? It needs to hum for years. Like a Mars rover, not a TikTok filter.
The manifesto flips the script. Make prompts explicit artifacts. ADRs for AI decisions. Tests that verify determinism. Reflections on automation in retros. It’s Agile, evolved — humans as conductors, AI as orchestra.
Skeptical? The original Agile snowbird scribbles birthed empires. This one’s crowdsourced, pain-driven. Star the repo. File issues. Co-author v1.0.
Look.
Teams ignoring this will stack bodies — codebases, that is. Adopters? They’ll ride AI’s wave to platforms we can’t yet dream.
Can the Manifesto Actually Stick in Real Sprints?
Short answer: hell yes, if you iterate it. The author’s teasing OpenAgile.ai next — a tool to weave these principles into workflows. But manifestos first. Principles over pixels.
I’ve shipping-scarred fingers itching to test this. Imagine: AI generates, humans trace and own. Domain-driven code that breathes your business. No more ‘works on my machine’ ghosts.
Pushback welcome. Is ‘human accountability always’ too rigid? Does it stifle solo indie hackers? Debate in the GitHub issues — that’s the point.
AI isn’t just a tool. It’s the new OS for creation. Vibe coding’s the beta bug. AVC? The stable release.
🧬 Related Insights
- Read more: Agent-Months: The Smoke and Mirrors of AI Agent Hype, Per Wes McKinney
- Read more: Hybrid Events: Blending Virtual Fire with In-Person Sparks in Open Source
Frequently Asked Questions
What is vibe coding?
Vibe coding is relying on AI tools like Cursor for rapid code generation without safeguards, leading to unmaintainable, buggy production systems over time.
How does the Agile Vibe Coding Manifesto fix AI coding problems?
It extends Agile principles with rules like traceable intent, human accountability, and versioned context to ensure AI outputs stay sustainable and understandable.
Will vibe coding kill traditional software engineering?
No — it amplifies it, but only with frameworks like AVC to prevent code graveyards and keep humans in the driver’s seat.
