> ## Documentation Index
> Fetch the complete documentation index at: https://docs.driven.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Skill Routing Logic

> How Driven automatically selects the right Skill for your question, how to influence the choice, and how to invoke a specific Skill manually when you want to.

Driven selects a [Skill](/concepts/skills) for your question automatically. This page explains how that works, how to steer it, and how to override it when you want a specific workflow.

## Automatic selection

By default, you do not name a Skill. You describe what you want, and Driven routes to the workflow that fits. The routing reads the intent of your prompt:

* A request to analyze one company routes to [Stock Analysis](/skills/stock-analysis)
* A description of screening criteria routes to [Stock Screener](/skills/stock-screener)
* A request to value a company routes to [Valuation Matrix](/skills/valuation-matrix)
* A request to compare companies routes to [Competitor Analysis](/skills/competitor-analysis)
* A broad, multi-source research question routes to [Deep Research](/skills/deep-research)
* A request about holdings, insiders, or flows routes to [Smart Money](/skills/smart-money)
* A request to review your holdings routes to [Portfolio Monitor](/skills/portfolio-monitor)
* A market overview routes to [Market Pulse](/skills/market-pulse); a sector scan to [Sector Radar](/skills/sector-radar)

## How to steer the choice

The clearer your intent, the more reliable the routing. Phrasing that names the task helps:

* "Value \[TICKER] and cross-check against peers" clearly signals valuation
* "Find companies with \[criteria]" clearly signals screening
* "Compare \[A] and \[B]" clearly signals comparison

Vague prompts ("tell me about \[TICKER]") can route to a general analysis when you wanted something specific. If you want valuation, say valuation.

## Naming a Skill explicitly

You can force a specific Skill by naming it in your prompt:

```text theme={null}
Use Deep Research to analyze [TICKER]'s competitive position and risks.
```

This bypasses the inference and runs the named Skill.

## Manual invocation

You can also invoke any Skill directly through the menu or the command line, independent of your prompt's wording. This is useful when:

* You know exactly which workflow you want
* You want to override the automatic choice
* You want to run a specific Skill on existing context

## When routing picks the "wrong" Skill

If the result is not the workflow you expected, you have two options: rephrase to signal the intent more clearly, or invoke the Skill you wanted explicitly. Either corrects it. You do not have to start a new conversation.

## Related

* [Skills concept](/concepts/skills) — what Skills are and how to invoke them
* [Skills reference](/skills/overview) — what each Skill does
* [Prompt library](/prompts/overview) — prompts written to route cleanly