Menu
Docs / Home

Project Intelligence lives here.

The docs carry the full hierarchy: Project Intelligence First, Workflow Continuity as the execution mechanism, outcome-weighted judgment as proof, and memory, MCP, APIs, Sandbox, orchestration, and agent workflows as supporting surfaces.

One-minute map

Architecture

The required path and optional layers: AI client, Hosted MCP or API, Project Intelligence, Companion, Orchestrator, Sandbox, and runtime targets.

Product model

Project Intelligence

The five-question model for AI agent work: what changed, why, impact, next step, and whether it is safe to proceed.

Workflow mechanism

Workflow continuity

How Project Intelligence turns into execution through Start Work Briefs, What Changed For Me, PR/MR Answer Packs, Decision Ledger, context authority, and collision risk.

Review workflow

PR/MR Answer Packs

The product artifact for pull requests and merge requests: what changed, impacted symbols, source links, stale warnings, tests, reviewers, and repository context.

Trust layer

Context Authority

The product contract for source_url, commit_sha, freshness, confidence, validation state, and stale warnings.

Developer reference

128-tool MCP reference

Hosted MCP tools for Project Intelligence retrieval, reviewed memory, code graph, workflow planning, authority metadata, and agent workflow integration.

Documentation stance

Nothing is hidden from the docs. The product can be simple at the top of the site and still expose every primitive, protocol, package, workflow, boundary, and advanced surface once a technical reader enters the documentation.

Complete catalog

Every public docs surface, grouped by why you came here.

Integrations

Client, editor, package, and SDK wiring. Hosted MCP stays the default path for agents.

API and authority

Reference surfaces for software integration, authentication, and hosted MCP contracts.

Sandbox, orchestration, and automation

The advanced surfaces that should be discoverable in docs without becoming homepage positioning.

Memory and agents

Reviewed memory and multi-agent concepts stay available as inputs to the Project Intelligence layer.