vibecodingisold
Pre-planning knowledge injection for AI-coded projects.
314 senior-engineering items across 7 roles, filtered through 6 project profiles, distributed via 5 tool integrations.
Why this exists
Two years into the AI-coding era, the same scaffold ships to production over and over. Auto-increment integer primary keys. JWT in browser localStorage. .env files committed to the repo. No idempotency on payment endpoints. No tenant_id column ("we'll multi-tenant later"). No rate limit on /login. No structured logs, no PII redaction, no trace propagation. No staging environment, no IaC, no rollback plan. Mocked tests that pass while the real DB migration takes the database offline.
Each defect is forgivable in isolation. Stacked, they're a six-week rebuild that the founder will do under time pressure when a paying customer reports a double-charge or an auditor asks for the audit log.
The diagnosis: AI-coding tools generate code well. They do not generate the right requirements. Senior engineers don't catch these defects because they write better prompts; they catch them because they bring seven different roles' worth of mental checklists to the design conversation. AI tools don't have those checklists, and the user usually doesn't either. vibecodingisold is the layer that closes the gap — it front-loads the missing 314 items as planning context before code is written, so the AI's plan accounts for them from Day 1.
Three specific historical postmortems anchor the bet:
- Knight Capital lost $440M in 45 minutes because they shipped without a runtime kill switch. (Architect Item 39 — kill-switch / runtime feature-flag commitment.)
- Cloudflare went down globally for 30 minutes in 2019 because one regex had no resource limit. (Backend Item 38 — ReDoS / catastrophic-backtracking protection.)
- GitLab almost lost everything in 2017 because all 5 backup mechanisms had silently failed. (Database Item 21 + DevOps Item 26 — verified-restore drill.)
Each of those is a single item in the canonical source. There are 311 more. We refuse to wait for the postmortem to learn what was missing.
What's in the box
- 314 Mandatory / Conditional / Flag-and-Defer items across 7 senior-engineering roles (architect, backend, database, security, devops, sre, qa)
- 6 project profiles that filter the relevant subset (b2b-saas-web 264, b2c-saas-web 248, marketplace 246, ai-wrapper 230, mobile-app 220, internal-tool 184)
- 6 worked examples — one per profile, all fictional products, all end-to-end (Step 1 — Step 7)
- 5 tool integrations — Cursor, Claude Code Skill, Cline, Aider, plain prompt-templates
- Claude Code plugin with
/vibecodingisold:init, /vibecodingisold:audit, /vibecodingisold:profile slash-commands
- 8 binding Numerical Contracts (Idempotency-Key TTL, liveness timeout, terminationGracePeriodSeconds floor, webhook timestamp tolerance, vuln-scan severity gate, secret rotation cadence, RPO/RTO floors)
- Methodology doc explaining how the knowledge base was built (Brain-Worker-Hybrid pattern)
- 6 machine-checkable validators that gate every change in CI
Quick start (Claude Code plugin)
One marketplace add, one install, one command:
/plugin marketplace add 0zoriginals/vibecodingisold
/plugin install vibecodingisold@vibecodingisold
/vibecodingisold:init
That generates PLAN-REQUIREMENTS.md for your detected project profile in your project root. Feed it to your planning agent (/gsd:plan-phase, Cursor /plan, Cline, Aider, or any LLM chat).
Three commands available after install:
