dsh-llm-guardian

When managing multiple LLM providers with DeepSeek Harness, users often struggle with opaque service health and difficult usage tracking, which can lead to unexpected downtime or budget overruns. The dsh-llm-guardian plugin is crafted to address these challenges by enhancing model management locally. Its core capabilities include real-time health checks, local token quota management, and a comprehensive usage dashboard. Health checks enable periodic or on-demand testing of provider endpoints, allowing users to spot issues early and maintain reliability; local counters track token usage precisely, with configurable quotas that block requests when limits are reached, helping to control costs effectively. Additionally, built-in query adapters support account balance checks for platforms like DeepSeek and Z.AI Coding Plans, while the dashboard aggregates session logs into 7, 30, or 90-day views, featuring model filtering, activity heatmaps, and trend analysis for intuitive insights. This plugin is ideal for active DeepSeek Harness users, particularly developers, researchers, or teams relying on LLMs for daily work. It streamlines multi-provider monitoring and ensures privacy through local data processing and security measures—such as not storing API keys or uploading data—making the experience more reliable, cost-effective, and secure.