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

# Signals overview

> Competitor & market intelligence, refreshed on your schedule.

Signals is a web-search-backed research pass that surfaces recent developments in your competitor set and market.

## The output

Each Signals run produces two things:

1. **Signal cards** — 3-8 items, each tagged with category (product, pricing, funding, hiring, GTM, market), competitor name, impact level (high / med / low), and a real source URL.
2. **Market digest** — 60-120 words of prose synthesising the run.

## How it works

1. You click **Run Research** (or **Refresh Research**) on the Signals page.
2. Xura sends your configured competitors, sources, and window (30/90/180/365 days) to the research agent.
3. The agent does `web_search` calls (capped at 5) to find recent stories.
4. Results are normalised, JSON-parsed, and returned to your Signals surface.
5. Persisted to your profile so the next visit shows the cached run.

## Sub-tabs

* **Market Intelligence** — signal cards + market digest.
* **Sources** — competitors, preferred domains, research model, your priorities.

## Empty results

If the agent finds nothing in your window, Xura auto-retries once with a wider window (2× your pick, capped at 365 days). Still empty results are **not cached** — the next page visit re-attempts fresh instead of showing "0 signals · updated 1 day ago" indefinitely.

## The 24-hour cooldown

Once a run produces real signals, the button locks for 24 hours. This exists because:

* Each run costs `web_search` calls + a large LLM turn (\~$0.10-$0.40 depending on model).
* Signals don't change materially inside 24h for most industries.
* Server enforces the cooldown too, not just client — bypassing storage clears won't help.

Empty runs don't count against the cooldown; you can retry immediately if a run finds nothing.