Philter

Philter is an API-based application from Philterd that identifies and manipulates sensitive information in text: redacting, anonymizing, encrypting, or replacing PII, PHI, and other sensitive data. It sits on top of Phileas, Philterd's open-source redaction engine, and adds a REST API plus a policy engine for managing filtering policies. For each entity type you choose how it is handled: mask it, encrypt it (including format-preserving encryption), replace it with a synthetic value, drop it, or pass it through. The policy engine also supports conditional redaction, such as only certain names or only credit cards that pass Luhn validation, plus custom entity types via dictionaries and identifier patterns.

Philter was released as open source under Apache 2.0 in July 2024 (version 2.6.0), though the project dates back to 2019, and it remains actively maintained with commits days old. It runs as a Java 17+ application via Docker, ships a web UI for policy management, and is also available as turnkey images on the AWS, Google Cloud, and Azure marketplaces. The Philterd toolkit around it includes Phileas libraries in Java, Python, .NET, and Go, a Philter AI Proxy that strips PII from prompts before they reach LLM providers, Phinder for discovery scanning, and Philter Scope for benchmarking redaction policies. Philter Desktop is a separate Windows app for redacting documents locally; its source is open under Apache 2.0 while the signed build is a commercial $100/user/year subscription. It is a solid pick for teams that need redaction inside their own VPC with full source access, though at 60 GitHub stars it is a niche project rather than a mainstream standard.

1. Clone https://github.com/philterd/philter (requires Java 17+ and Maven) or pull the Docker image.
2. Build with mvn clean install, then run with docker compose up.
3. Send text to the API: curl -k https://localhost:8080/api/filter --data "..." -H "Content-type: text/plain".
4. Open the UI at http://localhost:9000 to author and manage redaction policies.