Dynamic Skill Router
Paste the following prompt into your AI chat to install this skill:
Please follow https://skillhub.cn/install/skillhub.md and install @user_e8fae1af/skill-router-dynamic.
About this skill
The problem
When an agent workspace contains skills for multiple platforms and tasks, the model may not know which skill should handle the current request. Guessing by name can miss project-level skills, skipping frontmatter can misread triggers, and loading without verification can send work to the wrong capability.
How it works
skill-router-dynamic is a meta-routing skill that does not perform domain work itself. It routes requests to the correct SKILL.md. Its lifecycle includes:
- Entry Guard: decide whether the message is chat or a task, avoiding unnecessary routing.
- Scan: check project-level and user-level skill directories in priority order, then extract name, description, and triggers.
- Match: choose direct loading, present options, or report missing skills based on confidence.
- Verify: check whether the executed skill produced an on-intent result, and return to scanning on failure.
- Control Hook: keep the loaded skill in the same session, while new sessions re-enter routing.
Key outputs use the [skill-router] prefix so routing events can be searched in transcripts. Multi-skill patterns support sequential, parallel, and composed workflows, with context passed step by step.
Boundaries
It assumes readable SKILL.md files and conventional directories, so it is not a substitute for loose prompt files without frontmatter. Logging is opt-in and local by design. Routing quality still depends on how clearly candidate skills describe their purpose.
Use Cases
- A Cursor or Claude Code workspace has multiple SKILL.md files, and the agent must pick the right one.
- Project-level and user-level skills share the same name, so the router should scan project dirs first.
- A loaded skill returns an off-intent result, so the agent should verify, unload, and retry another candidate.
- Two skills, such as data cleaning and report generation, need to run sequentially with context passed between steps.
Best For
- AI agent engineers maintaining Claude Code or Cursor skill directories who need a unified routing entry point.
- Team leads defining multi-platform SKILL.md standards who need auditable routing and verification steps.
- Developers debugging skill-matching failures who need visible scan, match, and fallback records.
- Technical authors building local agent workspaces who need multi-skill orchestration patterns.
Related Skills
Provides Claw with character-library selection, switching, saving, and global SOUL.md style sync for role-based conversation.
An AIONE Agentic AI Infrastructure SDK wrapper for building production AI agents with memory, skills, workflows, and hooks.
A Python/TypeScript SDK wrapper for the DeepSeek-Reasonix native AI coding agent, with prefix-cache support.
A browser automation tool for analysts, operators, and developers that locates elements, fills forms, extracts structured content, and supports no-code scheduling and export.