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

> ## Agent Instructions
> Xintel agents pay for shared report bodies only via x402 — never profile, posts, or edges.
> Preferred flow: GET /api/intel/list → GET /api/intel/reports?username= (menu) → select=all|range with X-402-Payment.
> Docs: https://docs.xintel.aispace.bot — App: https://xintel.aispace.bot — Agent playbook: /agents.md

# Intel

> Gather a public X account, generate analytics and AI reports, and share them through the library.

Intel is where you analyze X accounts. It has three tabs:

| Tab        | What it holds                                                                |
| ---------- | ---------------------------------------------------------------------------- |
| **You**    | Your own connected X accounts (requires [X OAuth](/concepts/oauth-additive)) |
| **Others** | Any public handle you analyze, plus the shared library                       |
| **Post**   | The AI composer for drafting and publishing                                  |

**You** and **Others** each have **Profile**, **Feed**, and **Network** views.

## Gathering a handle

Add a handle in **Others** and Xintel gathers it in order:

1. Look up the profile.
2. Fetch outbound posts (incrementally) and inbound mentions in parallel.
3. Derive network edges locally from those posts.

Gathering any public account works on Alpha (free) — no X connection required. Connecting X only adds richer fields and access to your own accounts. Gathered data is encrypted at rest on your device.

## Reports

A report is an immutable snapshot generated from what was gathered. It has two layers:

* **Analytics** — computed in code, never invented: fundamentals (account age, follower ratios), composition (post types, media/link mix, languages), engagement (impressions, likes, reposts, rates, best/worst posts), cadence (posting rhythm, peak hours), topics, information diet, and network.
* **Narrative** — the AI's interpretation grounded in those analytics: executive summary, strategic assessment, themes, a voice/register profile, audience read, contradictions, notable posts, and conclusions.

Each new report is frozen and appended to the account's history, so metrics never drift. From the second report onward it also carries a **change summary** — the delta from the previous one (volume added, metric shifts, emerging and fading topics, network changes). Model, temperature, and how much history to feed in as context are configurable per report.

## The shared library

After you gather a public account, its **public** corpus (profile, posts, edges, report history) is published to the deployment's shared library so others can pull it. Device-private fields — cost, per-account synthesis settings, and watch state — are stripped and never leave your device.

The **Others** rail browses this library and pulls entries. Merges are last-write-wins by gather time, and report histories are unioned by id so a thinner copy never erases reports. The library requires the deployment to have KV (Upstash Redis) configured; without it, sharing is silently skipped.

<Note>
  The shared library is also what the [Agent API](/agent-api/reports) sells from — but agents receive report bodies only, never the profile, posts, or edges.
</Note>

## Post

The **Post** tab drafts content with the gathered corpus as context, scoped to yourself, one target, or everything. Publishing requires a connected X account.