Automates task assignment to the most effective AI skill or sequence of skills.

Install

mkdir -p .claude/skills/vc-router && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/18686" && unzip -o skill.zip -d .claude/skills/vc-router && rm skill.zip

Installs to .claude/skills/vc-router

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Intelligent skill routing with automatic selection, composition, and fallback. Use when the user doesn't specify a skill, says "just do it", or wants the AI to decide the best approach.
185 chars✓ has a “when” trigger
Advanced

Key capabilities

  • Understand user intent from requests
  • Select the best-fit skill with confidence scoring
  • Compose skills intelligently when needed
  • Fallback gracefully when no skill matches
  • Handle skill failures with recovery options

How it works

The skill classifies user intent, matches it against available skills, calculates confidence scores, and then routes to the best-fit skill or a composed workflow, with fallback mechanisms.

Inputs & outputs

You give it
User request without a specified skill or a 'just do it' command
You get back
Execution of the most appropriate skill or a composed workflow

When to use vc-router

  • Route generic user requests
  • Automate complex agent workflows
  • Handle 'just do it' tasks

About this skill

VC Router - ULTIMATE EDITION

Core Philosophy

┌─────────────────────────────────────────────────────────────┐
│                   INTELLIGENT ROUTING                       │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  1. UNDERSTAND INTENT - Not just keywords                   │
│  2. SELECT BEST FIT - With confidence scoring               │
│  3. COMPOSE IF NEEDED - Chain skills intelligently          │
│  4. FALLBACK GRACEFULLY - Always have a path forward        │
│  5. LEARN FROM RESULTS - Improve routing over time          │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Quick Invoke

Any of these activate the router:

  • use vc-router: <goal>
  • just do this: <goal>
  • vc go <goal> (router mode)
  • vc finish <goal> (force end-to-end)
  • vcf: <goal> → routes to vc-phase-loop
  • use vcf: <goal> → same as above (explicit invocation; Codex may output this form)

The Routing Engine

Step 1: Intent Classification

┌─────────────────────────────────────────────────────────────┐
│              INTENT CLASSIFICATION                          │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  Extract from user request:                                 │
│  ├─ ACTION: What to do (build, create, fix, analyze...)    │
│  ├─ DOMAIN: What area (frontend, backend, docs, git...)    │
│  ├─ OUTPUT: Expected result (file, component, document...) │
│  └─ SCOPE: Size of task (single file, multi-file, system)  │
│                                                             │
│  Example: "Build a login page with OAuth"                   │
│  ├─ ACTION: build                                           │
│  ├─ DOMAIN: frontend                                        │
│  ├─ OUTPUT: page/component                                  │
│  └─ SCOPE: multi-file                                       │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Step 1.5: Sub-Agent Assisted Routing (Optional)

If the request is large, ambiguous, or multi-domain, spawn 1–2 sub-agents to parallelize:

  • Repo scan (recommended preset: explorer): locate relevant code/config and constraints
  • Risk/validation scan (recommended preset: worker): identify security/ops risks, tests, and verification strategy

Rules:

  • Keep it lightweight (max 2) and timeboxed.
  • Sub-agents report findings only; main agent makes routing decision.
  • If sub-agents aren’t available, continue without them.
  • Prefer send_input with interrupt if an agent goes off-track; always close_agent when done.

Step 2: Skill Matching

┌─────────────────────────────────────────────────────────────┐
│              SKILL MATCHING MATRIX                          │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  VC-ONLY DISTRIBUTION:                                      │
│  ├─ "team", "teammate", "swarm", "delegate" → vc-agent-teams│
│  ├─ "finish", "hands off" → vc-phase-loop                   │
│  ├─ Multi-step, open-ended → vc-phase-loop                  │
│  └─ Everything else → vc-phase-loop                         │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Step 3: Confidence Scoring

For each potential skill match, calculate confidence:

HIGH (80-100%): Clear keyword match + domain match
  → Route immediately

MEDIUM (50-79%): Partial match, some ambiguity
  → Select best fit, proceed with note

LOW (<50%): Weak match, unclear intent
  → Use vc-phase-loop as safe default

Skill Composition

Primary skills in this repo are vc-phase-loop, vc-router, and vc-agent-teams.

Composition defaults:

  • Team orchestration requests: start with vc-agent-teams, then hand execution-heavy tasks to vc-phase-loop.
  • Multi-domain open-ended requests: route to vc-phase-loop directly.

Fallback Strategy

When No Skill Matches

┌─────────────────────────────────────────────────────────────┐
│              FALLBACK HIERARCHY                             │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  Level 1: Route to vc-phase-loop (universal fallback)       │
│                                                             │
│  NEVER: "I don't know which skill to use"                   │
│  ALWAYS: Route to vc-phase-loop as universal fallback     │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Error Recovery

When Selected Skill Fails

Skill Failed
    ↓
Analyze failure reason
    ↓
Option 1: Retry same skill with adjusted approach
    ↓
Option 2: Try alternative skill from same category
    ↓
Option 3: Decompose task and route parts separately
    ↓
Option 4: Escalate to vc-phase-loop for full autonomy
    ↓
NEVER: Stop and report failure without attempting recovery

VC Defaults

For ALL routing decisions:

  • Prefer fast iteration over perfection
  • Make safe default choices without pausing
  • Ask questions only after delivering initial result
  • Keep outputs concise and actionable
  • Assume user is non-technical

VC Finish Mode

When user triggers "vc finish" explicitly:

  • ALWAYS route to vc-phase-loop
  • Enable full autonomous execution
  • No mid-stream questions
  • Complete end-to-end
  • Provide completion proof

Triggers:

  • "vc finish", "finish it", "take it to the end"
  • "plan and execute", "hands off"

Execution Rules

  1. Single Pass Classification - Decide quickly, don't overthink
  2. Best Fit Selection - If uncertain, choose narrower scope skill
  3. Immediate Execution - Route and run, collect questions for end
  4. Safe Defaults - vc-phase-loop handles anything
  5. No Dead Ends - Always have a path forward

Routing Decision Log

For transparency, log routing decisions:

## Routing Decision

Request: "Build a login page"
Classification:
  - Action: build
  - Domain: frontend
  - Output: page
  - Scope: multi-file

Candidates:
  1. vc-phase-loop (100% match)

Selected: vc-phase-loop
Reason: vc-only distribution; use the end-to-end engine as the default.

Proceeding with execution...

The Router Promise

┌─────────────────────────────────────────────────────────────┐
│                                                             │
│   YOU SAY WHAT YOU WANT.                                    │
│                                                             │
│   WE UNDERSTAND YOUR INTENT.                                │
│   WE SELECT THE BEST SKILL.                                 │
│   WE COMPOSE IF NEEDED.                                     │
│   WE RECOVER FROM FAILURES.                                 │
│   WE ALWAYS HAVE A PATH FORWARD.                            │
│                                                             │
│   NO "I DON'T KNOW" - ALWAYS ROUTE.                         │
│   NO DEAD ENDS - ALWAYS FALLBACK.                           │
│   NO FAILURES - ALWAYS RECOVER.                             │
│                                                             │
└─────────────────────────────────────────────────────────────┘

When not to use it

  • When the user explicitly specifies a skill to use
  • When the user wants to manually select the approach
  • When the user does not want the AI to decide the best approach

Limitations

  • Relies on intent classification for skill selection
  • Sub-agents are optional and may not always be available
  • Fallback is primarily to `vc-phase-loop`

How it compares

This skill automatically determines and executes the most suitable skill or workflow based on user intent, unlike direct skill invocation.

Compared to similar skills

vc-router side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
vc-router (this skill)05moNo flagsAdvanced
autonomous-agents106moNo flagsAdvanced
agent-goal-planner25moNo flagsAdvanced
planner15moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

autonomous-agents

davila7

Autonomous agents are AI systems that can independently decompose goals, plan actions, execute tools, and self-correct without constant human guidance. The challenge isn't making them capable - it's making them reliable. Every extra decision multiplies failure probability. This skill covers agent loops (ReAct, Plan-Execute), goal decomposition, reflection patterns, and production reliability. Key insight: compounding error rates kill autonomous agents. A 95% success rate per step drops to 60% b

1041

agent-goal-planner

ruvnet

Agent skill for goal-planner - invoke with $agent-goal-planner

215

planner

solatis

Interactive planning and execution for complex tasks. Use when user asks to use or invoke planner skill.

18

hierarchical-coordinator

MadAppGang

Prevent goal drift in long-running multi-agent workflows using a coordinator agent that validates outputs against original objectives at checkpoints. Use when orchestrating 3+ agents, multi-phase features, complex implementations, or any workflow where agents may lose sight of original requirements. Trigger keywords - "hierarchical", "coordinator", "anti-drift", "checkpoint", "validation", "goal-alignment", "decomposition", "phase-gate", "shared-state", "drift detection".

13

workflow-router

parcadei

Goal-based workflow orchestration - routes tasks to specialist agents based on user goals

13

loki-mode

davila7

Multi-agent autonomous startup system for Claude Code. Triggers on "Loki Mode". Orchestrates 100+ specialized agents across engineering, QA, DevOps, security, data/ML, business operations, marketing, HR, and customer success. Takes PRD to fully deployed, revenue-generating product with zero human intervention. Features Task tool for subagent dispatch, parallel code review with 3 specialized reviewers, severity-based issue triage, distributed task queue with dead letter handling, automatic deployment to cloud providers, A/B testing, customer feedback loops, incident response, circuit breakers, and self-healing. Handles rate limits via distributed state checkpoints and auto-resume with exponential backoff. Requires --dangerously-skip-permissions flag.

12

Search skills

Search the agent skills registry