$ whoami
vibeeval — vibe coder from turkey
$ cat /etc/mission
built a 119-agent AI software team on Claude Code.
no custom models. no custom APIs. just engineering.
$ cat /proc/stats
AGENTS ............. 119 specialized roles
SKILLS ............. 202 reusable workflows
HOOKS .............. 48 automated gates
SELF-LEARNING ...... on errors → rules → better
| repo | about | |
|---|---|---|
drwxr-xr-x |
vibecosystem | the full AI team — 119 agents that plan, build, review, test |
drwxr-xr-x |
instinct | self-learning hooks — claude code gets smarter every session |
input: "build me X"
│
├─→ architect designs the approach
├─→ developers implement (frontend, backend, infra)
├─→ reviewers check quality + security
├─→ testers verify everything works
│
├─→ mistake found?
│ └─→ learned. added to rules. never happens again.
│
└─→ pattern detected across projects?
└─→ promoted to global knowledge.
PID SYSTEM STATUS
001 agent-swarm ██████████ multi-agent orchestration
002 dev-qa-loop ██████████ implement → review → retry
003 self-learning ██████████ errors → rules, auto
004 canavar-cross-training ██████████ one mistake trains all
005 instinct-pipeline ██████████ patterns mature over time
