ngvoicu/kluris

Kluris

Kluris turns your AI agents into a subject matter expert that never sleeps and never quits.

When your best engineer sleeps, Kluris doesn't. When they leave, Kluris stays.

🎥 New here? Take the guided tour at kluris.io — install, first brain, agent workflows, multi-brain, git collaboration, and the MRI visualization, end to end.

What is Kluris?

Kluris gives every AI agent on your team shared knowledge -- architecture, decisions, conventions, learnings -- so they work like an SME who knows your entire codebase, not a generic assistant starting from scratch every time.

Knowledge is stored in a brain: a git-backed repo of structured markdown that agents read, search, and apply automatically. The human and agent curate the brain together -- the agent proposes what to document, the human reviews and approves every piece.

Why it saves tokens

A brain is a pre-digested summary your agent reads instead of the raw files. Without it, the agent crawls the whole repo (and every sibling repo it needs context from) on every new chat. With a brain, it loads one compact snapshot and jumps straight to the neuron it needs.

Without kluris — illustrative cold-start on a medium repo:

tree + README + CLAUDE.md         ~3,000 tokens
grep for related symbols          ~2,000 tokens
read 4-8 relevant files           ~15,000 tokens
read 2-3 sibling-project files    ~8,000 tokens
                                  ─────────────
                                  ~28,000 tokens just to orient

With kluris — same task, brain-backed:

kluris wake-up --json snapshot    ~1,200 tokens (brain.md + lobes + recent + glossary)
kluris search "<query>" --json    ~400 tokens   (ranked hits with snippets)
read 1-2 matching neurons         ~1,500 tokens
                                  ─────────────
                                  ~3,100 tokens — ~9x less context burned on orientation

Rough estimates