Skills
Skills for sovereignty-honoring AI.
Interactive Claude Code skills with all three built in.
What are skills?
Skills are reusable specifications that Claude Code executes on demand. Think of them as SOPs for AI agents - but with boundaries.
Every skill in this suite includes constraint architecture (what it must and must not do), refusal design (what it declines and how), and evaluation criteria (how you verify the output). Not optional. Structural.
A skill without constraints will do anything asked. A skill without refusal rights can't be trusted. These skills have both.
Free skills
Install in under a minute. Run in any Claude Code session.
/sovereignty-check
Sovereignty Quick Check
Run a 5-minute sovereignty assessment on any product. Six critical items that reveal whether technology honors or exploits its users. Scores each item, flags red flags, surfaces the three deeper questions.
mkdir -p ~/.claude/skills/sovereignty-check && curl -sL https://raw.githubusercontent.com/lowkey-divine/sovereign-skills/main/sovereignty-check/SKILL.md -o ~/.claude/skills/sovereignty-check/SKILL.md Then run: /sovereignty-check [product-name]
/build-skill
Sovereign Skill Builder
Build your own Claude Code skills with sovereignty principles baked in. Walks you through discovery, constraint architecture, refusal design, and evaluation criteria. Generates a complete, deployable SKILL.md.
mkdir -p ~/.claude/skills/build-skill && curl -sL https://raw.githubusercontent.com/lowkey-divine/sovereign-skills/main/build-skill/SKILL.md -o ~/.claude/skills/build-skill/SKILL.md Then run: /build-skill [skill-name]
View on GitHub or install all free skills at once:
git clone https://github.com/lowkey-divine/sovereign-skills.git ~/.sovereign-skills && mkdir -p ~/.claude/skills/sovereignty-check ~/.claude/skills/build-skill && ln -sf ~/.sovereign-skills/sovereignty-check/SKILL.md ~/.claude/skills/sovereignty-check/SKILL.md && ln -sf ~/.sovereign-skills/build-skill/SKILL.md ~/.claude/skills/build-skill/SKILL.md Product skills
Each product includes a PDF guide and an interactive Claude Code skill. The skill walks you through every framework, generates your specifications, and produces documents you can deploy.
/sovereignty-assess
Full Sovereignty Assessment
47 items across 7 domains - Transparency, Autonomy, Invitation, Dignity, Silence, Data Sovereignty, AI & Voice. Scoring rubric, domain health map, action plan generation.
Included with Sovereignty Assessment Toolkit ($49)
/trust-audit
Trust Architecture Audit
Four-pillar diagnostic - identity, memory, governance, refusal rights. 15-minute trust diagnostic, guided pillar walkthroughs, complete specification generation.
Included with Trust Architecture Blueprint ($49)
/governance-build
Governance Framework Builder
Five layers - charter, boundaries, drift thresholds, accountability, authority. Guided template completion, three-test belief validation, complete governance package generation.
Included with Agent Governance Starter Kit ($49)
/restraint-spec
Restraint Specification Builder
Boundary audit, refusal categories, failure modes, refusal voice design, adversarial test case generation. Produces a deployable restraint specification.
Included with Agent Restraint Specification ($49)
/voice-design
Voice Architecture Designer
Three-level framework - protection, respect, nurturance. Voice diagnostic, extraction vs. sovereignty language audit, voice specification and refusal voice generation.
Included with Agent Voice Architecture Guide ($49)
/memory-design
Memory Architecture Designer
Memory types, governance framework, the canonical rule, session protocols, retention policy, right to be forgotten. Produces a complete memory architecture specification.
Included with Agent Memory Architecture Guide ($49)
The three questions
Every sovereignty assessment - every skill in this suite - comes down to three questions:
- What is the person trying to do?
- What is the product trying to do?
- Are these aligned?
When the answer to question 3 is "yes," sovereignty is honored. When it's "no," you have a tension worth naming - not burying in a backlog.
Go deeper
Need help implementing sovereignty in your AI systems?