A Claude Code skill that evaluates whether your product follows the principles of Dignified Technology — tools that protect and elevate the creative, expressive, and irreplaceable dimensions of human work.
Based on the essay "What Does a Tool Owe You?", which argues that a tool owes you your own capability back, amplified — not a replacement for your thinking, but a deepening of it.
Six principles, each scored A through F:
| Principle | Core Question |
|---|---|
| Authorship Preservation | Would the user sign their name on the output without hesitation? |
| Creative Range | Does the tool expand creative range, or flatten it? |
| Voice Amplification | Does it amplify the user's voice, or replace it? |
| Process Involvement | Did the tool ask questions, or just produce output? |
| Depth of Exploration | Did the user reach an insight they wouldn't have found alone? |
| Input Agency | Does the user govern what reaches them? |
Plus a Values-as-Spec assessment: has the team explicitly defined what values the product protects, how users are involved, and what the product refuses to automate?
The audit uses a diagnosis-first hybrid approach:
- Codebase Diagnosis — Reads your project (CLAUDE.md, README, source code, docs) to form preliminary assessments
- Targeted Interview — Asks only about gaps the diagnosis couldn't resolve from code
- Values-as-Spec Assessment — Three questions about product values (always asked)
- Scorecard — Letter grades per principle, overall Dignity Grade, and actionable recommendations
One-liner (clones the repo and installs automatically):
curl -sSL https://raw.githubusercontent.com/xaelophone/dignified-technology/main/install.sh | bashOr clone first, then install:
git clone https://github.com/xaelophone/dignified-technology.git
cd dignified-technology
./install.shRestart Claude Code after installing.
In any project directory with Claude Code:
/dignified-technology
The skill will read your codebase, ask targeted questions, and produce a scorecard.
See the full Hermes audit for a real example. Summary:
Overall Dignity Grade: A
Authorship Preservation — A: AI never writes prose — it reads, questions, and highlights.
Creative Range — B: 8 highlight types surface multiple angles; no structured divergent phase.
Voice Amplification — B: Voice highlights flag style deviations using prior essays.
Process Involvement — A: Multi-step loop with user decisions at every point.
Depth of Exploration — B: Challenges arguments reactively. Reactive by design.
Input Agency — A: All AI features are opt-in. Default state is quiet.
The full essay is included at essay/what-does-a-tool-owe-you.md for reference.
This is an open framework. If you've run the audit on your product and have suggestions for improving the principles, rubric, or interview flow, open an issue or PR.
MIT