VoidAccess
提出一个问题,VoidAccess 便并行搜索 12+ 个 Tor 搜索引擎、GitHub 与安全订阅源,然后返回可供 STIX、MISP 或 Sigma 使用的行动者、钱包、泄露凭证和 IOCs。Open so
AI 翻译
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证Hide private details in a document before you share it with an AI assistant, then put them back. Offline, on-device, free. By Philanthropel.
翻译准备中
取自仓库 README
正在加载完整 README。
查看完整 README ↗You have a donor letter, a spreadsheet, or a report. You'd like ChatGPT, Claude, or Copilot to help you rewrite it, summarise it, or tidy it up. But you can't paste it in, because it's full of things you must not share.
Penname fixes that in four steps:
What counts as a private detail? Names of donors, staff and trustees. Organisations. Email addresses, phone numbers and postal addresses. Gift amounts and giving histories. Wealth and capacity ratings. Fund, campaign and appeal codes. Constituent IDs.
| You can open | What you get back |
|---|---|
Word (.docx) | A Word file |
Excel (.xlsx) | An Excel file |
| CSV spreadsheets | A CSV file |
Plain text (.txt, .md) | A text file |
| A text copy you can paste anywhere |
Penname also knows the export formats of the systems you already use, including Raiser's Edge NXT, Salesforce NPSP, DonorPerfect, Bloomerang and Little Green Light.
Scanned PDFs won't work yet. If your PDF is a photograph or scan of a page rather than real text, Penname can't read it. Copy the text into a Word or text file first.
This is the part that matters most, so here it is plainly:
The only moment anything leaves your computer is when you choose to paste the safe copy into your AI assistant. And by then the private details are gone.
Penname helps you do one thing that nearly every privacy law in the world asks for: share less personal data than you otherwise would.
Under UK and EU GDPR this is called pseudonymisation, and the law names it directly — Article 4(5) defines it, and Article 32 lists it as an appropriate security measure. Elsewhere it goes by other names, but the principle is the same: don't hand over personal details you don't need to hand over.
| Where you work | The rules that apply | How Penname helps |
|---|---|---|
| United Kingdom | UK GDPR, Data Protection Act 2018 | Pseudonymisation is a recognised safeguard |
| European Union / EEA | GDPR | Named in Article 32 as a security measure |
| United States | State privacy laws — CCPA/CPRA in California, plus Virginia, Colorado, Connecticut, Texas, Utah and more | Less personal information disclosed to a third-party service |
| Canada | PIPEDA | Limits what is disclosed to others |
| Australia | Privacy Act 1988 — especially APP 8, on sending data overseas | An AI assistant is usually an overseas disclosure |
| India | Digital Personal Data Protection Act 2023 | Reduces the personal data processed |
| South Africa | POPIA | Supports minimality and security safeguards |
| Brazil | LGPD | Supports data minimisation |
| Kenya / Nigeria | Data Protection Act 2019 / NDPA 2023 | Reduces what is shared with processors |
Please read this part
Penname does not make you compliant, and it does not make your data anonymous. Because the real details can be restored, the safe copy still counts as personal data under GDPR and most similar laws. What Penname does is reduce how much personal data you share, which is a genuine and recognised safeguard — not a get-out.
You are still responsible for reviewing what Penname produces and for meeting the rules that apply to you. Penname is a tool, not legal advice. If your work involves health, financial, or children's data, check with whoever advises you before using any AI assistant.
Please read this before you download. Penname is free, open-source software, so it isn't registered with Apple or Microsoft the way paid software is. Your computer will warn you the first time. That's expected, and here's exactly how to get past it. You only do this once.
Click Download for Mac below. You'll get a file called Penname.dmg.
Open it. A window appears with the Penname icon and an Applications folder. Drag Penname onto Applications.
Now one extra step. Open the Terminal app — press ⌘ +
Space, type Terminal, press Enter. Copy the line
below, paste it into the black window, and press Enter:
xattr -dr com.apple.quarantine /Applications/Penname.app
It will look like nothing happened. That's correct. You can close Terminal.
Open Penname from your Applications folder. It will open normally from now on.
What did that command do? It told your Mac "I downloaded this on purpose." Nothing more. It doesn't change Penname or give it any new powers.
Right-clicking and choosing Open used to work instead of step 3. Apple removed that in macOS 15, so the Terminal step is now the reliable way.
Penname-Setup.exe.Software companies pay Apple and Microsoft every year to have their apps recognised. Penname is given away free, so it doesn't carry that registration yet, and your computer says so rather than staying silent about it.
The warning means "nobody has paid to vouch for this" — not "this is dangerous." If you'd rather be certain, every release publishes a SHA-256 checksum and a build provenance attestation proving the file was built from this public code, untouched by anyone. See Verifying your download.
Mac users: Penname currently requires a Mac with Apple Silicon (M1, M2, M3 or M4). To check, click the menu → About This Mac. If it says "Intel", Penname won't run on your machine yet — please tell us so we can prioritise it.
Drag your file onto the window, or click Choose a file. Penname reads it and looks for private details.
The first document you open each day takes a little longer, because Penname is starting up its language tools. You'll see a progress bar while it works.
This is the screen that matters. Penname shows every detail it found, what kind of thing it thinks it is, how sure it is, and the stand-in it suggests.
No tool finds everything, which is why this step exists and why Penname will never skip it for you. Read your document once more before you share it.
Penname saves two files:
.pnmap) — encrypted, and it stays on your computer. Keep
it, because it's what puts the real details back in step 4.Do your work in your AI assistant as normal. When you have its reply, copy it, come back to Penname, and paste it into Take pen names off.
Penname puts the real names and numbers back and saves the result as a file on your computer. The restored text is never shown to the AI, and never sent anywhere.
That's the whole loop: Open → Check → Save & share → Bring the details back.
Penname is open source (Apache-2.0). The engine is a pure Python library with the desktop app, a command-line tool, and an optional MCP server all sitting on top.
uv sync # install dependencies (Python 3.11+)
uv run pytest # run the test suite — round-trip integrity is the gate
uv run python -m penname.cli.main pseudonymize letter.txt --mapping letter.pnmap
uv run python -m penname.cli.main reverse response.txt --mapping letter.pnmap -o restored.txt
If your project folder path contains a space,
uv's editable install writes a.pththe interpreter skips, so thepennameconsole command may report "No module named 'penname'". Runuv run python -m penname …instead, or clone to a path without spaces. Installed releases and CI are unaffected.
Detection combines Microsoft Presidio (regex patterns for emails, phones, amounts, IDs, fund codes) with spaCy for named entities. An optional GLiNER backend adds higher-recall detection for names, organisations, and places.
The downloadable apps ship Presidio + spaCy only — bundling GLiNER pulls in PyTorch, which pushes the app past GitHub's 2 GB release limit. GLiNER is a source-install extra; the engine uses it when present and falls back to Presidio + spaCy when it isn't:
uv sync --extra gliner # install GLiNER + torch
uv run python -m penname.core.detect.gliner_model # fetch the model (one-time)
The model is only ever downloaded by that fetch step; at runtime it loads from
local files with the network disabled. (GLiNER can't be installed on Intel Macs —
onnxruntime no longer ships Intel-Mac wheels — so on those the app runs
Presidio + spaCy only.)
Penname ships an MCP server so a local AI agent can pseudonymize documents and take pen names off replies. It is disabled by default:
python -m penname.mcp.config enable # opt in (writes ~/.penname/settings.json)
penname-mcp # starts the server over stdio
It exposes two tools: pseudonymize_document (returns file paths and counts,
never raw content) and reverse_to_file (writes the restored text to a local
file and returns only a success flag and the path — restored donor data is
never returned into the model's context).
Every release lists a SHA-256 checksum next to each file and a build provenance attestation proving the file was built by this repository's release workflow. To check your download matches:
# macOS
shasum -a 256 Penname-macos.zip
# Windows (PowerShell)
Get-FileHash Penname-windows.zip -Algorithm SHA256
Compare the result to the .sha256 file published with the release.
Penname is made and maintained by Philanthropel, which builds bespoke agentic systems for the third sector, and is given freely to the nonprofit and philanthropy community.
© 2026 Philanthropel Limited.
LICENSE). You may use, modify, and redistribute it under those
terms.NOTICE). The open-source
license covers the source code only — it does not grant any right to use these
names or logos, so please don't ship forks or other products under the Penname
or Philanthropel name.Penname runs entirely on your computer and makes no network connections, so no information you process with it is ever sent to Philanthropel or anyone else.