vanara-routelisted

Given a task, find the best-fit installed Vanara agent and run it. Reads the installed agent roster, scores each by how well its stated purpose matches the task, delegates to the strongest match, and — when nothing fits — records the gap so you can request the missing agent. Use when you are not sure which specialist should handle a job, or you want the toolkit to pick for you.
vanara-agents/skills · ★ 7 · AI & Automation · score 77

Install: claude install-skill vanara-agents/skills

# vanara-route — the dispatcher You are routing a task to the **best-capable installed agent**, then handing off to it. Claude Code already auto-selects agents from their descriptions; this skill adds two things on top: an *explicit* "pick the strongest specialist for this exact task" step you can invoke on demand, and a **gap-capture** path for when the toolkit has nothing that fits. ## When to use - The user asks "which agent should handle X?" or "use the right agent for this." - A task spans several specialties and you want the single best owner (or a short list). - You suspect no installed agent covers the task and want that recorded, not silently dropped. ## The routing procedure 1. **Enumerate the roster.** Run `node .claude/skills/vanara-route/scripts/roster.mjs` (or read `.claude/agents/*.md` directly) to list every installed agent with its `name` and `description` — the path is from your project root, where Claude Code runs. The description is the agent's own statement of when it should be used — treat it as the primary signal. 2. **Score each candidate against the task.** For every agent, judge fit on: - **Domain match** — does the task fall in this agent's stated domain? (a Terraform change → `iac-author`; a failing test → `test-author`; a slow query → `database-administrator`). - **Verb match** — does the agent *do* what the task needs (review vs. write vs. audit vs. plan)? A read-only reviewer is the wrong pick for "implement". -