noirdoc

noirdoc is a skill for Claude Code from noirdoc-ai/mask-claude-plugin. It costs 144 tokens per session (3,144 once invoked), scanned A, original, MIT.

A privacy-focused workflow for working with local documents that may contain personal information, such as contracts, invoices, employment records, or medical text. It hides identifying details while the document is being processed and restores them for the final result.

In plain words
What is it for?
Use it to read, summarize, translate, analyze, or edit qualifying local documents. It can also support redaction, pseudonymization, and GDPR-related document handling.
Why use it?
It reduces the chance that names, account numbers, or other personal details enter the working context unnecessarily. It is meant for sensitive documents and privacy-related tasks.

Skill for Claude Code

Written for Claude Code: ${CLAUDE_PLUGIN_ROOT} variable. Also seen: mentions subagents; names the NotebookEdit tool; mentions Claude Code.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the noirdoc plugin — 1 skill, 4 commands, 1 hook shipped together

Good fit Use it to read, summarize, translate, analyze, or edit qualifying local documents. It can also support redaction, pseudonymization, and GDPR-related document handling.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add noirdoc-ai/mask-claude-plugin
Claude Code
/plugin install noirdoc

Made for: Claude Code.

Or install noirdoc, the plugin that ships this one along with the rest of its 1 skill, 4 commands, 1 hook.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for noirdoc

README.md
[![agentmods](https://agentmods.dev/badge/skills/noirdoc-ai/mask-claude-plugin/noirdoc/github.svg)](https://agentmods.dev/skills/noirdoc-ai/mask-claude-plugin/noirdoc)
Your own site
<a href="https://agentmods.dev/skills/noirdoc-ai/mask-claude-plugin/noirdoc"><img src="https://agentmods.dev/badge/skills/noirdoc-ai/mask-claude-plugin/noirdoc/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for noirdoc

Your own site · 80×15
<a href="https://agentmods.dev/skills/noirdoc-ai/mask-claude-plugin/noirdoc"><img src="https://agentmods.dev/badge/skills/noirdoc-ai/mask-claude-plugin/noirdoc.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 144 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,144 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00144 $0.03144
Opus 5 $0.00072 $0.01572
Sonnet 5 $0.00029 $0.00629
Haiku 4.5 $0.00014 $0.00314

Measured 10d ago against content hash 1ac25a1544bb, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

noirdoc scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

noirdoc/skills/noirdoc/SKILL.md · 159 lines

How it starts

The opening of the file, as written. The whole thing — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.

noirdoc — privacy-preserving document workflow

This skill lets you work with sensitive local documents without pulling raw PII into your context. It orchestrates the noirdoc CLI: redact on the way in, reveal on the way out, so the user sees real names in your final output while your working context only holds placeholders like <<PERSON_1>> and <<IBAN_CODE_1>>.

When this applies

Use this skill when:

  • The user asks you to read / summarize / translate / analyze / edit a document under ./incoming/, ./clients/, ./contracts/, or any file matching *.contract.*, *.nda.*, *.patient.*.
  • A PreToolUse block comes back with "noirdoc-protected" in the message.
  • The user says anything about "redact", "pseudonymize", "DSGVO", "PII", "GDPR", "anonymize names", "noirdoc", or German legal/HR/medical documents.
  • The user drops a PDF/DOCX/XLSX of a contract, employment agreement, invoice, or similar into the workspace and asks you to work with it.

Don't use this skill when:

  • The user's question is purely about code, configuration, or non-document files (source files, JSON configs that are clearly not PII).
  • The file is already a redacted copy (look for .noirdoc/cache/ prefix, or placeholders like <<PERSON_1>> in a preview).
  • The user has explicitly told you the document contains no personal data and asked you to skip redaction — respect that, don't argue.

First-run setup

Before anything else, check whether this workspace is already set up:

test -f .noirdoc/config.toml && echo SETUP || echo NOT_SETUP

If NOT_SETUP, walk the user through setup before redacting anything:

  1. Check noirdoc is installed. noirdoc --version — if the command is not found, ask the user:

    "This workspace isn't set up for noirdoc yet. I can install it for you — the full German-first ensemble needs about 560 MB of disk (spaCy + Flair + GLiNER weights). Install noirdoc[full] now? (y/n)"

    On yes: pip install 'noirdoc[full]' && noirdoc models pull On no: stop. Tell the user you can't proceed safely without the CLI, and offer to continue without redaction only if they explicitly accept that the document will enter your context in the clear.

Read the full file on GitHub · 159 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 10d ago First seen · 159 lines · 144 tokens per session scan A 1ac25a1544bb

Subscribe to this mod's changes

noirdoc is a skill published in the GitHub repository noirdoc-ai/mask-claude-plugin (2 stars, last pushed 3mo ago), licensed MIT. It adds 144 tokens to every session and 3,144 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.