Daimon: Local Proxy Redacts Sensitive Prompts Before External Large Language Model Inference
Daimon is an open-source local proxy that intercepts LLM requests to sanitize credentials and private identifiers before sending them to external providers. Once the remote response returns, Daimon restores the original local data automatically.
Impact: Medium
Why it matters
You can now leverage cloud models like OpenAI Codex while preventing API keys, passwords, and internal identifiers from leaking over network requests.
TL;DR
- 01Daimon strips sensitive strings locally using configurable regex rules before calling external LLMs.
- 02Responses are reconstructed on the local machine with original values without leaking private text.
- 03Built-in debug UI at localhost:3010/debug allows developers to inspect what gets sent over the wire.
Key facts
- Chat Interface
- http://localhost:3010
- Debug Inspector
- http://localhost:3010/debug
- Config Dashboard
- http://localhost:3010/config
Local Prompt Sanitization Architecture
Daimon deploys locally to protect credentials, private tokens, internal IDs, and document bases before payloads leave your environment. Instead of transmitting raw text to remote providers, Daimon runs local regex pattern matching to replace sensitive strings with isolated tags.
Local Reconstruction and Endpoint Compatibility
When the remote LLM returns its generation, Daimon executes a reverse substitution pipeline. The raw sensitive strings never touch remote servers. Daimon exposes an OpenAI-compatible endpoint, enabling direct integration into existing client workflows and agent environments.
Administration and Debugging Tools
The proxy provides three core local web interfaces on port 3010:
http://localhost:3010: Main chat interface supporting protected, local, and remote execution modes.http://localhost:3010/debug: Real-time inspection dashboard showing raw inputs, sanitized outgoing payloads, and final reconstructed outputs.http://localhost:3010/config: Administrative panel to configure regex rules, prompt structures, private RAG document ingestion, and external OAuth linkages.
Try it in 2 minutes
# Access Daimon local web interfaces
# Chat interface:
open http://localhost:3010
# Pipeline inspection and sanitization debugger:
open http://localhost:3010/debug
# Model, prompt, and regex configuration panel:
open http://localhost:3010/configbash
✓ When to use
- Routing proprietary codebases or private text through cloud LLM APIs without exposing raw secrets.
- Building local-first RAG pipelines where full document bases must remain protected on-premises.
- Debugging and auditing exactly what data an agent sends to external OpenAI-compatible endpoints.
What to do today
- Clone the Daimon repository and link external LLM credentials via the openai-oauth connector.
- Configure custom regex filters and prompt rules in the management UI at http://localhost:3010/config.
- Validate sanitized outgoing payloads using the inspection dashboard at http://localhost:3010/debug.
Sources