{"id":"f08501b0-7a12-4625-ba7a-598c3a59d850","slug":"privacygate","kind":"skill","name":"privacygate","summary":"A Claude Skill for anonymising personal, sensitive, health (PHI), financial, and confidential information before it's shared with an AI system.","owner":"IslandNC","repoUrl":"https://github.com/IslandNC/privacygate","license":"MIT","licenseTier":"permissive","trust":"community","stars":0,"match":null,"fewStars":true,"format":"skill","createdAt":"2026-08-02T06:40:20.000Z","addedAt":"2026-10-21T13:23:07.548Z","logoUrl":"https://avatars.githubusercontent.com/u/96751409?u=3123aafd9f6043fc6f3217ac10e970f3dc8a0266&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["data-analytics"],"tags":["ai-safety","anonymisation","claude","data-protection","de-identification","phi","pii","privacy"],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-23T05:51:03.748Z","sample":false,"url":"https://conduits.dev/tools/privacygate","language":"","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# PrivacyGate\n\nA Claude Skill for anonymising personal, sensitive, health (PHI), financial, and confidential\ninformation in a document before it's summarised, analysed, indexed, archived, or shared with\nan AI system.\n\nIt produces two outputs:\n\n- **Mapping register** — real values mapped to fictitious placeholders (`Person_001`,\n  `Organisation_001`, etc.), kept private, used only for controlled re-identification.\n- **AI-safe working copy** — the same document with every sensitive item replaced by its\n  placeholder, safe to process, share, or store.\n\n## Why\n\nMost \"just anonymise it\" prompts fall apart because they don't guarantee the same real value\nmaps to the same placeholder every time, and they don't separate \"what to keep private\" from\n\"what's safe to share.\" PrivacyGate treats this as a deterministic mapping problem: build a\nregister first, then work exclusively from placeholders.\n\n## Status\n\nPersonal project, built and maintained independently. Not affiliated with, endorsed by, or\nrepresenting any employer or organisation. See [DISCLAIMER.md](DISCLAIMER.md) before you use\nthis for anything with real legal or compliance weight.\n\n## How it works\n\n1. Identify direct, indirect, employment, health, financial, technical, and special-category\n   identifiers in the source content.\n2. Flag combinations that could re-identify someone even without a direct identifier.\n3. Classify each item (security classification, privacy category, confidence, recommended\n   action: Remove / Generalise / Pseudonymise / Anonymise / Retain).\n4. Build a deterministic, one-to-one mapping register — fictitious placeholders only.\n5. Produce the AI-safe working copy.\n6. Run a second pass to catch anything missed, plus a reconstruction test and residual risk note.\n7. From that point on, only placeholders are used — real values aren't restated unless you\n   explicitly ask for them back.\n\nFull workflow detail: [`privacygate/SKILL.md`](privacygate/SKILL.md).\n\n## Install\n\n1. Download [`privacygate.skill`](privacygate.skill) from this repo (or clone it and package\n   the `privacygate/` folder yourself — see below).\n2. In Claude, upload/install the `.skill` file. It installs to your skill library and does\n   **not** run automatically.\n3. Invoke it explicitly when you want it — e.g.:\n   > \"Use PrivacyGate to anonymise this email thread before I paste it into a report.\"\n   > \"Run a privacy review on this CV.\"\n\nIt will not trigger on casual mentions of \"privacy\" or \"sensitive\" in unrelated conversation —\nthe trigger is intentionally narrow. See [`docs/usage.md`](docs/usage.md) for worked examples.\n\n## Repackaging from source\n\nIf you modify `privacygate/SKILL.md` or the reference files, repackage with the official\nAnthropic skill-creator packaging script (bundled with Claude's skill-creator tooling), or\nsimply zip the `privacygate/` folder — a `.skill` file is a zip containing the skill directory.\n\n## Repo structure\n\n```\nprivacygate/\n├── SKILL.md                          # workflow instructions (what triggers it, the 11 steps)\n├── references/\n│   └── taxonomy.md                   # identifier categories, loaded only when the skill runs\n└── assets/\n    ├── mapping-register-template.md  # register column structure (fictitious example values only)\n    └── output-structure.md           # optional full 12-section formal report structure\ndocs/\n└── usage.md                          # worked example, all fictitious data\nDISCLAIMER.md\nLICENSE\n```\n\n## Limitations\n\n- This is a workflow aid, not a certified compliance tool. It does not replace a formal Privacy\n  Impact Assessment, legal advice, or sign-off from an actual Data Protection Officer.\n- Output quality depends on the model actually running it — always review the mapping register\n  and working copy yourself before relying on either.\n- It has no persistent storage of its own across separate conversations; the mapping register\n  is only consistent within a single session unless you save and reuse it yourself.\n\n## License\n\nMIT — see [LICENSE](LICENSE).\n"}