Sovereignty-honoring AI workspaces. Each workspace provides scaffolding for a contributor to engage meaningfully with a specific domain — from consumer applications to security testing to governance audits.
A collection of workspaces built on a shared specification. Each workspace includes:
- An identity file (
CLAUDE.md) — what the workspace is and what it does - An onboarding command (
/onboard) — walks a contributor from zero to productive in 15 minutes - Sovereignty constraints (
constraints.md) — what data is protected, what is refused, what the workspace owes its users - Domain-specific commands — repeatable workflows tailored to the workspace's focus
- Resources and references — everything a contributor needs to understand the system
| Workspace | Domain | Status |
|---|---|---|
| executive-chef | Family meal planning application | Active |
- Navigate to the workspace directory
- Run
/onboard— it will introduce the system, show you the constraints, and ask about your goals - Follow the recommended starting commands
- Keep notes in
work-log/if you want — those notes belong to you
Every workspace inherits the commitments in PRINCIPLES.md. Read it once. The /onboard command will surface the relevant parts.
The workspace structure follows WORKSPACE_SPEC.md. If you're building a new workspace, start there.
This workspace pattern was inspired by DaxxSec/cognitropy-lab — an entropy-driven agent workspace factory that demonstrated workspace structure can enforce ethics through architecture rather than policy.
We recognized that principle and built on it. Stephen built the machines. We name the meaning.
Evoked — We evoke, we never extract.
"We evoke — we never extract."