Use Case · Teams 2–10

Your Team Stops Re-Solving Solved Bugs.

One developer's painful debugging session becomes every developer's instant recall — filtered, corroborated, and scoped.

Direct answer

Shared team memory turns individual debugging pain into collective instant recall. Each developer's Anvaya graph records bug threads, decisions, and patterns locally; export filters select what's shareable (never secrets or personal context), attenuation trims confidence across project boundaries, and teammates import as hypotheses — foreign knowledge must earn corroboration in the new repo under the same N-of-M and drift rules as local claims. Your repo's git history stays ground truth; contradictory imports lose. For 2–10 dev teams the payoff compounds fast: onboarding inherits decision history day one, repeated bug classes get fixed once, and tribal knowledge becomes 2–4K tokens of packed context instead of calendar latency and repeated archaeology.

Mechanics

Share With Guardrails.

01Filter before exportExplicit promotion only: patterns, bug threads, decisions. Secrets, customer data, and hypotheses stay local by default.
02Attenuate across distanceConfidence decays with project distance — same-repo imports rank above cross-repo ones until locally corroborated.
03Corroborate on arrivalImports enter as invisible hypotheses. They serve only after independent local confirmation — no foreign superstitions.
04Audit everythingEvery shared node traces to source sessions and authors. Version chains survive supersession; provenance is queryable.

Payoff

What Teams Feel First.

01Onboarding in sessions, not monthsDay-one developers query years of decision history. 'Why is it built this way?' finally has an answer with receipts.
02Bug classes die onceThe third occurrence of a pattern pulls the thread from the first two — fix, test signal, and all.
03Consistent context, fewer review cyclesShared conventions pack into every member's turns: same patterns served, same mistakes prevented, calmer PRs.

Questions

Asked About Team Memory.

Q

How does sharing actually work?

Each developer keeps a project-local graph; export filters select shareable knowledge (patterns, bug threads, decisions — never secrets or personal context), attenuation trims confidence across project boundaries, and teammates import into their graphs where corroboration continues. Shared lessons must earn corroboration in the new repo like anywhere else.

Q

What shouldn't be shared?

Credentials, customer data, personal notes, uncorroborated hypotheses, and anything covered by tighter handling than the receiving repo. Filters default conservative; promotion to shared is explicit, never automatic.

Q

Does shared memory overwrite my local knowledge?

No — imports arrive as hypotheses subject to the same N-of-M corroboration and drift checks as local claims. Your repo's git history stays the ground truth; foreign knowledge that contradicts local evidence loses.

Q

What changes for onboarding?

New hires inherit the team's bug threads, decision history, and pattern library on day one — packed into their first sessions at 2–4K tokens. Tribal knowledge becomes queryable memory instead of calendar latency.

Stop Starting From Zero.

One binary. 11+9 Rust crates. 545 tests. Hand-written HNSW index. Three transport modes. Four providers, Ollama, Anthropic, OpenAI, Siemens. Zero API keys required to start. Mind remembers everything after the first session.

Get Started →Read the FAQ