Skip to content
Back to skills

Privacy Accounting

ASecurity

"Routes TensorFlow Privacy users who want privacy budgets,

  • 247 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 9, 2026
toolsapi

Works with

  • cli
  • api

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned September 9, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill privacy-accounting --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Privacy Accounting?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Privacy Accounting
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-privacy-accounting/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-privacy-accounting)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: privacy-accounting
description: "Routes TensorFlow Privacy users who want privacy budgets,
  epsilon/delta statements, or the repo's privacy accounting CLIs."
disable-model-invocation: true
metadata:
  disco-role: operating
license: Apache 2.0
---

# Privacy accounting

Use this sub-skill when the user wants to compute privacy budgets, find a noise multiplier, or explain the accounting assumptions behind a DP-SGD run.

## Trigger phrases

- "compute epsilon"
- "find the noise multiplier"
- "DP-SGD privacy statement"
- "privacy budget"
- "accountant type"
- "user-level privacy"
- "tree aggregation accounting"

## What this sub-skill covers

- `compute_dp_sgd_privacy` / `compute_dp_sgd_privacy_statement`
- `compute_noise_from_budget`
- the `RDP` and `PLD` accountant choices
- tree-aggregation privacy accounting helpers
- user-level privacy notes when `max_examples_per_user` matters

## What it does not cover

- training-loop integration -> `../training/`
- `DPQuery` internals -> `../queries/`
- attack analysis -> `../privacy-tests/`
- fast clipping internals -> `../fast-clipping/`

## Read this before you act

- `references/api-reference.md` for the verified function signatures and the CLI flag map.
- `references/troubleshooting.md` for missing-flag, delta, user-level, and accountant-selection failures.
- `../../references/install-and-scope.md` for the minimum runtime.

## Typical workflow

1. Determine whether the user needs an example-level or user-level statement.
2. Choose `RDP` unless the user explicitly wants `PLD`.
3. Use the bundled CLI helper when a command-line answer is enough.
4. Use the API reference when the user wants to embed the calculation in code.
5. If the user wants a tree-aggregation variant, read the tree-accounting notes before answering.

## Bundled helpers

- `scripts/compute_dp_sgd_privacy.py` mirrors the privacy-statement CLI with the same core parameters.
- `scripts/compute_noise_from_budget.py` mirrors the inverse search CLI that solves for a noise multiplier.

## Cross-links

- `../training/` usually feeds the values that this sub-skill consumes.
- `../queries/` owns the lower-level query math that eventually feeds some accounting paths.

Files in this skill

  • SKILL.md2.2 KB
  • references/api-reference.md2 KB
  • references/troubleshooting.md2.1 KB
  • scripts/compute_dp_sgd_privacy.py2.3 KB
  • scripts/compute_noise_from_budget.py1.7 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.