Get started
Core concepts in five minutes.
Everything Stable Baseline does sits on a small, stable set of building blocks. Once you understand these, the rest of the platform follows naturally.
Workspace
A workspace is the top-level boundary — usually your company or team. Every person, project, document, diagram, and plan belongs to exactly one workspace. Permissions, billing, and activity logs all scope to the workspace.
Project
A workspace has many projects. A project is a logical grouping for a product, service, or initiative — anything with its own docs, diagrams, and work plan. Most teams end up with 3–10 projects per workspace.
Document
Documents are where knowledge lives — specs, runbooks, PRDs, onboarding guides, API references, postmortems. The editor handles rich prose, lists, tables, code blocks, callouts, and inline diagrams. Every edit is versioned automatically.
Friendly IDs
Diagram
Diagrams are first-class — not attachments. Stable Baseline supports 40+ types (architecture, sequence, ERD, BPMN, Gantt, charts, mind maps, and many more) and renders them inline next to the prose that explains them. Every diagram is versioned just like the surrounding document.
Plan, phases, tasks
A plan is a timeline for a project. Plans contain phases (Discovery, Design, Build, Launch…), each with their own colour, date range, and tasks. Tasks carry an assignee, status, priority, and percent complete, and render as bars on the Gantt view.
Improvements
Improvements are the backlog. Bugs, tech debt, documentation gaps, risks, feature requests — anything that might become work later. Each one has a status, priority, type, and owner, and can be linked to docs, diagrams, or promoted into a plan task when it's ready to build.
Compliance
Stable Baseline tracks which of your documents satisfy which compliance requirements (SOC 2, ISO 27001, HIPAA, and more). Missing coverage is surfaced as findings that flow into the improvements backlog — so compliance stays a side-effect of doing the work, not a separate project.
Agents
An agent is any AI assistant you connect to your workspace (Claude Code, Cursor, Copilot in VS Code, and many more). Agents are members of the workspace with scoped permissions — read-only, edit, or full access — and everything they do is versioned and attributed to them, just like a human teammate.
The knowledge graph
Docs link to diagrams, plans link to docs, improvements link to evidence, and compliance findings link back to the docs that satisfy them. All of that forms a single knowledge graph under the hood — so opening any object shows you everything else it touches.