Keep Aadhaar, PAN and other Indian personal data out of LLM prompts - detection, a policy proxy for Claude Code, Copilot and Ollama, and an audit log. R and Python.https://dwarpal.github.io/credgate/ https://github.com/dwarpal/credgateen Tue, 01 Jan 1980 00:00:00 -0000 Tue, 01 Jan 1980 00:00:00 -0000 1440 MkDocs RSS plugin - v1.17.9 None https://dwarpal.github.io/credgate/ Bharat Sikka Claude Code Proxy Running Claude Code on Indian customer data, with every PAN, Aadhaar and mobile number tokenised or redacted before it leaves the machine, and an audit trail to prove it. https://dwarpal.github.io/credgate/blog/claude-code-without-a-pan/ Tue, 01 Jan 1980 00:00:00 +0000 credgatehttps://dwarpal.github.io/credgate/blog/claude-code-without-a-pan/ Bharat Sikka Aishwarya Bais Benchmarks Detection credgate, MaskFlow, Microsoft Presidio and Kavach PII on two public Indian PII datasets and credgate's own corpus - where each one is strong, where credgate trails, and what running other people's data taught us. https://dwarpal.github.io/credgate/blog/indian-pii-benchmarks/ Tue, 01 Jan 1980 00:00:00 +0000 credgatehttps://dwarpal.github.io/credgate/blog/indian-pii-benchmarks/ Bharat Sikka Aishwarya Bais Detection R Find Aadhaar, PAN, mobile numbers and other Indian personal data in R text and data frames, and redact or tokenise it before a prompt reaches a language model - with ellmer, Ollama or any provider. https://dwarpal.github.io/credgate/blog/redacting-indian-pii-in-r/ Tue, 01 Jan 1980 00:00:00 +0000 credgatehttps://dwarpal.github.io/credgate/blog/redacting-indian-pii-in-r/ Bharat Sikka Aishwarya Bais Policy editor R A Shiny app that lets InfoSec decide which language models may see which personal data, test the decision before saving it, and keep a tamper-evident record of every change - without editing YAML. https://dwarpal.github.io/credgate/blog/shiny-policy-editor/ Tue, 01 Jan 1980 00:00:00 +0000 credgatehttps://dwarpal.github.io/credgate/blog/shiny-policy-editor/