Library/melange_heimdallForked

zetic-ai/melange_heimdall

melange_heimdall

An on-device guardian for LLM API calls — safety, privacy, and cost optimization that never leaves the phone

View on GitHub↗Upstream zetic-ai/melange_heimdall↗

Builder

zetic-ai

zetic-ai • individual

Stars

Using upstream star count

Forks

Using upstream fork count

Open Issues

Activity Score

0/100

0 commits in 30d

Created

Mar 6, 2026

Project creation date

README Summary

Melange Heimdall is an on-device guardian system for LLM API calls that provides safety filtering, privacy protection, and cost optimization without sending data to external servers. The system runs entirely on mobile devices to ensure user data never leaves the phone while managing AI interactions. It acts as a protective layer between users and LLM services to enhance security and control costs.

AI Dev Skills

Unmapped

LLM API IntegrationOn-device Safety FilteringPrivacy-preserving Machine LearningCost Optimization for LLM APIsMobile AI/MLContent ModerationAPI Interception and ProxyingOn-device Machine Learning InferenceLLM Safety and Content ModerationPrivacy-preserving AIMobile AI/ML IntegrationReal-time Request FilteringEdge AI ArchitectureAdversarial Input Detection

Taxonomy

AI Trends

On-device AI AI Safety Edge AI Small Language Models Privacy-preserving AI Edge Machine Learning Privacy-preserving Machine Learning

Deployment Context

Edge/Mobile On-device On-premise

Industries

Developer Tools Healthcare FinTech Legal Tech

Modalities

Text

Skill Areas

LLM API Integration On-device Safety Filtering Privacy-preserving Machine Learning Cost Optimization for LLM APIs Mobile AI/ML Content Moderation API Interception and Proxying On-device Machine Learning Inference LLM Safety and Content Moderation Privacy-preserving AI Mobile AI/ML Integration Real-time Request Filtering Edge AI Architecture Adversarial Input Detection

Use Cases

LLM API Safety Filtering Cost Optimization for Mobile Apps On-device Content Moderation Privacy Protection for LLM Interactions API Call Monitoring and Logging On-device content safety filtering for mobile LLM applications Cost optimization through intelligent API request routing Privacy-first LLM integration without server transmission Real-time prompt injection and adversarial input detection Sensitive data masking before API submission

Recent Activity

Updated 1 months ago

7 Days

30 Days

90 Days

Quality

prototype

Quality: medium
Maturity: prototype

PM Skills

Developer Platform

Languages

Swift100.0%

Timeline

Project created: Mar 6, 2026
Forked: Mar 26, 2026
Your last push: 1 months ago
Upstream last push: 1 months ago
Tracked since: Mar 10, 2026

Similar Repos

pgvector cosine similarity · $0

Loading…