yjh051108/dsh-routing-suite

dsh-routing-suite — injector + router-standard kit: install the runtime injector first, then the task-aware reasoning-mode router preset (measured P1-P23).

View on GitHub ↗Jump to charts ↓

Summary Information

Updated 13 minutes ago
Added to GitGenius on September 1st, 2026
Created on August 14th, 2026
Open Issues & Pull Requests: 40 (+0)
GitHub issues: Enabled
Number of forks: 152
Total Stargazers: 7,125 (+0)
Total Subscribers: 18 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 11.5 hours
Mean response time: 3.5 days
90th percentile: 15.3 days
Tracked items: 53

How this project is maintained

Around half of the issues opened in the past year never receive a reply. 97% of open issues come from outside the core team, so the backlog reflects real-world use rather than internal planning. Only 11% of issues opened in the past year have been closed. Three people close 78% of everything that gets resolved.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 36
New in 7 days: 12
Closed in 7 days: 22
Avg open age: 11 days
Stale 30+ days: 0
Stale 90+ days: 0

Recent activity

Opened in 7 days: 12
Closed in 7 days: 21
Comments in 7 days: 40
Events in 7 days: 75

Top labels

  • bug (8)

Detailed Description

dsh-routing-suite is a runtime injector and router preset kit for AI agents built on the DSH framework.

The tool addresses the problem of routing AI agent reasoning modes and task awareness by combining two components: a runtime injector that enables hot-reloading and dynamic tool management without restarting the runtime, and a task-aware router preset that classifies agent behavior into reasoning modes (spec, react, mixed, weak) based on model capabilities and task requirements. The router preset uses persona-based routing with fixed guidance injection after each user message to maintain task focus and prevent topic drift, achieving reported convergence rates of 96% with zero-shot task completion improved from 0% to 100%.

Developers should adopt this tool if they are working with the DSH framework and need fine-grained control over how AI agents reason through different task types. The suite is designed for projects requiring task-aware routing where different models benefit from different reasoning strategies—for example, using spec-mode (planning-focused) prompting for Pro models and neutral classification for Flash models. The injector component provides development utilities for inspecting and adjusting router behavior at runtime.

The project maintains both the injector and preset as directories within a single repository while preserving upstream independent repositories for separate release cycles. The preset includes multiple routing strategies: router-standard with persona classification and complete sections, and router-spec prioritizing deep reasoning, with router-pro planned for future release.