lemonadesaltbagel

Veil — AI skill for Claude Code

AI community

Local privacy proxy for AI coding agents.

How to install Veil

This entry records only its repository, not the path inside it, so there is no exact command to give. Open lemonadesaltbagel/veil and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Veil does

Local privacy proxy for AI coding agents. The model sees consistent pseudonyms instead of real names, emails and secrets from your databases, logs and fixtures; commands the agent runs get the real values back locally. Works with Claude Code, Codex CLI, Bedrock, Vertex and OpenAI-compatible clients.

Alternatives in AI

Xerj — Local-first search engine and MCP server for AI agents 12.5k ★
DontFeedTheAI — Transparent anonymization proxy for AI-assisted pentesting 654 ★
Memory Forge Rs — Stop resetting satisfying AI chats — edit the memory instead 474 ★

README

veil

**Privacy middleware for AI coding agents.** The model sees consistent pseudonyms; your machine keeps the real values.

what the agent read locally          what the LLM receives                what you see in the answer
---------------------------          ---------------------                ---------------------------
Leonie Köhler                        First_Ttnx Last_Ttnx                 Leonie Köhler
leonekohler@surfeu.de                user_ttnx@example.test               leonekohler@surfeu.de
card=4111 1111 1111 1111             card=card_y2vb                       card=4111 1111 1111 1111
/Users/alice/code/shop               /Users/user_pjgu/code/shop           /Users/alice/code/shop

Coding agents send everything they read to the model provider: database query results, logs, fixtures, stack traces, even your home-directory name. veil is a local proxy that pseudonymizes all of it before it leaves your machine. Commands the agent runs get the real values back locally, so it can still query your real dev database and fix real bugs.

**Status: alpha.** It works end to end with Claude Code, Codex CLI and OpenAI-compatible clients, and it has been tested against real models. It hasn't been used in production yet. See [Limitations](#limitations).

How it works

 coding agent ──► veil (127.0.0.1:8787) ──► model provider
   (real data)    redact ▸ guard ▸ forward      (pseudonyms only)
              ◄── rehydrate per tool policy ◄──

Seeding (the core idea). veil learns the real values from your own data sources: Postgres, MySQL/MariaDB, MongoDB, Redis, SQLite, and fixture files. It classifies fields from native PII labels, column and JSON-path names, and value sampling. It then matches those exact values wherever they appear (logs, free text, JSON, code). Regex and NER guess; seeding knows.
Coherent pseudonyms. Pseudonyms are deterministic HMAC tokens. Everything about one person shares one code (`user_