rizzo-pii is a local tool that finds personal information in Italian legal documents and replaces it with reversible placeholders before the text is sent to a language model. It is for law firms, accountants, notaries, and other GDPR-bound users who need to work with sensitive documents while keeping the original data on their device.
Borrowing it
Nothing to install: this file belongs to Rizzo-AI-Academy/rizzo-pii. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/Rizzo-AI-Academy/rizzo-pii/main/CLAUDE.mdgit clone --depth 1 https://github.com/Rizzo-AI-Academy/rizzo-piiWrote 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.
[](https://agentmods.dev/instructions/rizzo-ai-academy/rizzo-pii/claude-md)<a href="https://agentmods.dev/instructions/rizzo-ai-academy/rizzo-pii/claude-md"><img src="https://agentmods.dev/badge/instructions/rizzo-ai-academy/rizzo-pii/claude-md/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.
<a href="https://agentmods.dev/instructions/rizzo-ai-academy/rizzo-pii/claude-md"><img src="https://agentmods.dev/badge/instructions/rizzo-ai-academy/rizzo-pii/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.06047 | $0.06047 |
| Opus 5.5 | $0.02419 | $0.02419 |
| Sonnet 5.5 | $0.01209 | $0.01209 |
| Haiku 4.5 | $0.00605 | $0.00605 |
Grade A, and why
rizzo-pii CLAUDE.md 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 2mo 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.
How it starts
The opening of the file, as written. The whole thing — 285 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Guida per Claude Code (claude.ai/code) quando lavora in questa repo. Panoramica e struttura delle cartelle in README.md. Documenti di dettaglio: docs/TASSONOMIA_TAG.md (i 22 tag), docs/DATASET.md (composizione completa di train/validation) e docs/TRAINING.md (unione dei due dataset HF, accortezze sui tag, iperparametri).
Cos'è questo progetto
Pipeline per addestrare un modello mmBERT (jhu-clsp/mmBERT-base) a fare token
classification di PII, con focus su testi legali italiani (atti, contratti, sentenze)
ma con training multilingue. Obiettivo finale: anonimizzare documenti in locale
prima di mandarli a LLM closed (anonymize → placeholder + dizionario reversibile locale →
API → ricostruzione), per studi legali / compliance GDPR.
Scelta di mmBERT (encoder multilingue, architettura ModernBERT, context nativa 8192) e non ModernBERT vanilla perché quest'ultimo è quasi solo inglese.
Ambiente — vincoli critici e non ovvi
GPU RTX 5060 Ti (Blackwell, sm_120) su Windows; Python in D:\programmi\python.
Questi punti fanno fallire tutto se ignorati:
- torch DEVE essere build cu128:
torch 2.11.0+cu128(dahttps://download.pytorch.org/whl/cu128). Le build cpu/cu121 non supportano sm_120 →torch.cuda.is_available()False o crash. - torchvision/torchaudio vanno disinstallati se a versioni vecchie: rompono l'import di
transformers con
operator torchvision::nms does not exist. Non servono qui. - accelerate ≥ 1.14 (transformers 4.57 chiama
unwrap_model(keep_torch_compile=...)). - Windows: nel Trainer
dataloader_num_workers=0(altrimentiRuntimeError ... bootstrapping). - seqeval non si compila (bug setuptools_scm) → metriche entity-level (P/R/F1) calcolate a mano dentro gli script, nessuna dipendenza.
- Dipendenze extra installate:
wandb(tracking) epython-dotenv(legge.env). - Log da PowerShell: redirezioni
>eTee-Objectscrivono UTF-16 → leggerli conGet-Content/python, NON con lo strumento Read. Gli script forzanosys.stdoutUTF-8.
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.
- 2mo ago First seen · 285 lines · 6,047 tokens per session scan A fb982ee50c99
rizzo-pii CLAUDE.md is an instructions file published in the GitHub repository Rizzo-AI-Academy/rizzo-pii (1,246 stars, last pushed 15d ago), licensed MIT. It adds 6,047 tokens to every session, about $0.0242 per session on Opus 5.5. 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 2026-08-30.
rizzo-pii CLAUDE.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| rizzo-pii CLAUDE.mdRizzo-AI-Academy/rizzo-pii | Instructions file | 1.2k | 15d ago | MIT | A | 6,047 | Claude Code |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | today | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |