skill-router

Selects the smallest sufficient set of audit, security, and reliability skills for a task.

Updated Aug 29, 2026
One-click install
npx skills add https://github.com/1arley/volibear --skill skill-router-1arley
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-router
Source: https://github.com/1arley/volibear/tree/main/.opencode/skills/skill-router
Command: npx skills add https://github.com/1arley/volibear --skill skill-router-1arley

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Non-trivial audits and reviews often suffer from skill explosion (running every adjacent skill) or under-routing (missing a verifier for critical mechanisms). This Skill routes a task to the minimum set of skills needed to falsify its relevant assumptions, using a structured catalog, risk budgets, and overlap penalties instead of guesswork. ## Core Features & Use Cases - Deterministic routing: Scores candidate skills using trigger matches, domain match, risk floors, composition bonuses, and overlap/cost penalties defined in a central catalog. - Risk-based budgets: Enforces skill count budgets by risk level (trivial 0-1 up to critical 3-6) and orders selections by role (generator, investigator, verifier, reviewer, researcher). - Traceable decisions: Produces a structured dispatch report with selected skills, scores, evidence, near misses, and research routing decisions. - Use Case: Given a high-risk task mixing security and reliability concerns, classify the risk and domain, read the catalog, and output an ordered list of two to four justified skills plus up to three near misses. ## Quick Start Ask the agent to route the task "review the payment retry logic for concurrency and permission issues" and produce the skill dispatch report.

Frequently Asked Questions about skill-router

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I choose which skills to run for a code audit?▼

Classify the task's risk level and dominant domain, then match candidate skills against catalog triggers and required signals. Score candidates with domain, cost, composition, and overlap factors, and keep only the highest-scoring skills within the risk budget.

How many skills should a high-risk task use?▼

A high-risk task has a budget of two to four skills, while critical tasks allow three to six. Exceeding the budget requires a concrete justification of new coverage or confirmation that no selected skill already provides.

What is the difference between under-routing and over-routing?▼

Under-routing means a critical task gets one broad generator with no verifier to confirm mechanisms like retries or ownership. Over-routing means one matching keyword pulls in every adjacent skill; remove each skill and check whether any falsifiable assumption is lost.

Does negative context like "no permissions" trigger security skills?▼

No. Negated terms such as "no permissions, no shared state" must not activate security or reliability skills. Routing decisions rely on context and required signals, not isolated keywords.

When should research routing be included in a dispatch?▼

Route to research when external uncertainty is the primary problem and the task cannot be resolved within the project. The dispatch report states none, proportional, or full research routing along with the sources or skills needed.