Getting started

Pages

Page	Covers
Company search	A complete authorized retrieval path
Company brains	Eight scenarios and a persistent, permission-aware answer application
Governed knowledge	Typed evidence and revisioned artifacts
Agent knowledge	Completed activity to reviewable proposals
Conversation knowledge	Settled conversations to searchable, evidence-bound episodes
Algorithm choices	Independent algorithms, workloads, fidelity, and pinned project citations
Complete module and API index	Source-derived definitions and public methods across the package
Install	Base package, optional integrations, injected dependencies
Architecture	Layer ownership, data flow, framework and storage boundaries
Maturity	Stability and support labels
Identity migration	Canonical document IDs and configured source fingerprints
Shared dependency updates	Atom reuse, completed receipts, and change propagation
Incremental maintenance	Selection, indexed planning, stable groups, and reversible aggregates

Package boundary

Supplied by Mari	Injected by the application
Types, planning functions, metrics, traces	Models, HTTP clients, databases, authorization, schedulers
Reference in-memory algorithms	Production scaling and transaction guarantees

• Build company search
• Company brains
• Build governed knowledge
• Build knowledge from completed agent work
• Install
• Algorithm choices
• Complete module and API index
• Architecture
• Maturity labels
• Identity migration
• Shared dependency updates
• Incremental knowledge maintenance