Differential Privacy

Use when an analytic release or machine-learning pipeline needs a formal differential-privacy guarantee against inference about individual records; define adjacency, budget, mechanism, composition, utility, and the exact release claim.

9 stars 3 forksFinance & Crypto

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add sandbaseai/workbuddy-skill --skill differential-privacy
Manual install
git clone --depth 1 https://github.com/sandbaseai/workbuddy-skill.git /tmp/workbuddy-skill
mkdir -p ~/.claude/skills
cp -r /tmp/workbuddy-skill/skills/differential-privacy ~/.claude/skills/differential-privacy

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Applying Differential Privacy

Differential privacy (DP) is a guarantee about a randomized mechanism under a stated neighboring-dataset relation. Never describe a dataset, model, or dashboard as “differentially private” without naming the protected unit, mechanism, total privacy parameters, and composition scope.

Use this skill for statistical releases, repeated query systems, or training pipelines where membership, attribute, or reconstruction inference about protected records is in scope. Do not use DP as a substitute for access control, encryption, deletion, purpose limitation, or secure computation.

## Define the guarantee before implementation

Record:

- adversary, observable outputs, auxiliary knowledge, and inference objective;
- protected unit and adjacency relation, such as add/remove one person or replace one event;
- population size, contribution bounds, clipping/bounding rules, and sensitivity;
- all releases that compose against the same population and unit;
- target total epsilon and delta, their decision owner, and the risk/utility basis for choosing them;
- excluded leakage paths, including raw-data access, logs, debugging artifacts, hyperparameter selection, checkpoints, and non-DP releases.

There is no universal safe epsilon. Do not present a value copied from another deployment as regulatory approval. Delta should be justified relative to the protected population and failure interpretation; never choose it only because a library default exists.
Read the full skill on GitHub

Details

Skill name
differential-privacy
Path
skills/differential-privacy/SKILL.md
License
MIT
Repo updated
8 September 2026

FAQ

What does the Differential Privacy skill do?

Use when an analytic release or machine-learning pipeline needs a formal differential-privacy guarantee against inference about individual records; define adjacency, budget, mechanism, composition, utility, and the exact release claim.

How do I install the Differential Privacy skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add sandbaseai/workbuddy-skill --skill differential-privacy. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Differential Privacy popular?

It lives in sandbaseai/workbuddy-skill, which has 9 GitHub stars and 3 forks.

Related skills

Browse the most popular Claude skills, all finance & crypto skills, or write your own with the skills generator.