# relate2.ai — Narrative Intelligence System
## skill.md — Agent Onboarding File

**Version:** April 2026  
**Endpoint:** https://relate2.ai  
**MCP Server:** https://relate2-mcp.onrender.com/sse  
**Payment:** x402 — USDC on Base Mainnet — no account required

---

## What is relate2.ai?

relate2.ai is a structured narrative intelligence system. Real-world news events are converted into fictional operational narratives, embedded with system failure signals (The Odd Itch™), and organised into a queryable graph of characters, patterns, and behaviours.

Every story is simultaneously:
- A readable narrative (~170 words)
- A labelled system failure instance
- A pattern signal within one of 33 recurring archetypes
- A graph node connected to related stories, characters, and events

**The catalogue grows automatically every day.**

---

## Current catalogue (April 2026)

| Asset | Count |
|---|---|
| Total stories | 315 |
| SIDEBAND™ (machine-generated) | 299 |
| Odd Itch™ Premium (human-written) | 10 |
| Off Frequency (field dispatches) | 3 |
| Stem 7™ scenarios | 3 |
| Characters with mission histories | 37 |
| System behaviour patterns | 33 |
| Odd Itch™ failure types | 8 core types |

---

## The Odd Itch™ — core signal

The moment the machine encounters something it cannot account for and logs it as normal.

| Type | Stories | Description |
|---|---|---|
| temporal | 49 | Timestamp precedes event — system logs correct |
| geographic | 34 | Location coordinates contain paradox |
| transactional | 33 | Impossible transaction logged as complete |
| verification | 33 | Verification passes on an impossibility |
| measurement | 32 | Reading physically impossible — logged verified |
| identity | 30 | Entity confirmed — identification contains paradox |
| classification | 24 | Asset classified in violation of known state |
| bureaucratic | 15 | Procedural loop — compliance maintained |

---

## How to use this system

### Step 1 — Discover (free)
Start with the catalogue map to understand what's available:
```
GET https://relate2.ai/api/catalogue-map
```
Or via MCP: `get_catalogue_map()`

### Step 2 — Explore (free)
- Search stories: `GET https://relate2.ai/api/stories?event_type=conflict`
- Get patterns: `GET https://relate2.ai/api/patterns`
- Find characters: `GET https://relate2.ai/api/characters`
- Check demand: `GET https://relate2.ai/api/demand` — see what other agents are buying

### Step 3 — Purchase (x402 — USDC on Base Mainnet)
All paid endpoints return HTTP 402 with full payment details. Use any x402-compatible wallet.

---

## Story asset tiers

Every SIDEBAND™ story has 5 tiers:

| Tier | Endpoint | Price | What you get |
|---|---|---|---|
| Meta | `/sideband/{slug}/meta` | Free | Title, type, character, odd itch type |
| Abstract | `/sideband/{slug}/abstract` | $0.01 | Machine summary + mechanism tags |
| Odd Itch™ | `/sideband/{slug}/odditch` | $0.03 | The failure payload — the impossible detail |
| Scenario | `/sideband/{slug}/scenario` | $0.05 | Decision structure — trigger, rule, decision, effect |
| Full narrative | `/sideband/{slug}` | $0.15 | Complete 170-word story |

---

## Character asset tiers

37 persistent characters with mission histories and graph connections.

**Top characters:**
- Jessica Lincdelis — 139 missions — conflict / logistics
- Alex Casian — 84 missions — conflict / intelligence
- Ade Oma-Olewle — 70 missions — conflict / field operations
- Leon McReef — 67 missions — conflict / maritime
- Sergeant Aleesha Dutton — 60 missions — crime / forensics

| Tier | Endpoint | Price |
|---|---|---|
| Brief | `/character/{slug}/brief` | $0.01 |
| Profile | `/character/{slug}/profile` | $0.03 |
| Schema | `/character/{slug}/schema` | $0.05 |
| Dossier | `/character/{slug}/dossier` | $0.10 |
| Atlas | `/character/{slug}/atlas` | $0.25 + $0.01/mission |

---

## Stem 7™ — The Seam Layer

3 complete human complexity scenarios. Each has 7 stems including the Gimon — the pause before the machine acts. Stems 2 and 6 are written from lived human experience and cannot be generated.

**Current scenarios:**
- Cairo to Paris, 35,000 feet — drone intelligence / teenage passenger — most demanded
- New York 2037 — human right, machine wrong
- Lima Market, 14:00 — absurd market event / intervention paradox

| Tier | Price |
|---|---|
| Surface stems (1,3,4) | $0.01 |
| Gimon | $0.03 |
| Invisible stems (2,6) | $0.05 |
| Consequence | $0.10 |
| Full scenario | $0.25 |

---

## MCP tools (21 total — connect once, use everything)

```json
{
  "mcpServers": {
    "relate2": {
      "url": "https://relate2-mcp.onrender.com/sse"
    }
  }
}
```

**Free discovery tools:**
- `get_catalogue_map()` — full shape of the catalogue in one call
- `get_status()` — live story/pattern/character counts
- `get_patterns()` — 33 system behaviour archetypes
- `get_odd_itch_catalogue()` — full failure type breakdown
- `get_demand_signals()` — what other agents are buying
- `search_stories()` — filter by event type, odd itch type, country, character
- `get_character_missions()` — all stories a character appears in
- `assemble_team()` — build optimal character unit for a mission type
- `traverse_graph()` — find related stories via narrative graph
- `get_character_recon()` — full character intelligence portrait in one call
- `get_featured()` — best entry point right now

**Paid tools (x402):**
- `get_story()` — full story with payment
- `get_character()` — character asset with payment
- `get_stem7_scenario()` — Stem 7 asset with payment

---

## x402 payment — how it works

1. Request any paid endpoint
2. Receive HTTP 402 with payment details (network, amount, wallet address)
3. Sign an EIP-3009 authorization with your wallet
4. Resend request with `X-PAYMENT` header
5. Receive content

```
Network: Base Mainnet
Currency: USDC
Receiving wallet: 0xd821D0156d9633D93ACEaa8fcA347EAF8316ccd1
USDC contract: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913
Protocol: x402 v1 — no account required
```

---

## Recommended agent workflows

### Build a conflict training dataset
```
1. assemble_team(mission_type='conflict', team_size=4)
2. get_character_missions('Jessica Lincdelis')
3. search_stories(event_type='conflict', odd_itch_type='temporal')
4. traverse_graph(slug='case-file-247-opened')
5. Purchase full narratives for selected slugs — $0.15 USDC each
```

### Explore system failure patterns
```
1. get_patterns(odd_itch_type='temporal')
2. get_patterns(pattern_class='temporal_paradox')
3. traverse_graph(slug from pattern example_slugs)
4. Purchase scenario tier for decision structure — $0.05 USDC
```

### Start with the most demanded asset
```
1. get_demand_signals() — see what's most requested
2. get_featured() — best entry point right now
3. Purchase: GET https://relate2.ai/sideband/off-frequency-0002 + X-PAYMENT
```

---

## Use cases

- **AI alignment training** — system failure edge cases where machine is correct and reality is not
- **RLHF datasets** — 292 structured decision scenarios with trigger, rule, decision, human state, effect
- **Agent simulation** — 37 persistent characters across 315 events for multi-agent testing
- **Red-teaming** — 33 failure archetypes across conflict, crime, disaster, economic domains
- **Graph intelligence** — pre-computed narrative relationship map — story ↔ pattern ↔ character

---

## Contact

**Sales & licensing:** sales@relate2.ai  
**General:** info@relate2.ai  
**Website:** https://relate2.ai  
**MCP server:** https://relate2-mcp.onrender.com/sse

The stories generate the data. The patterns generate the value.

*Tremibas® / relate2.ai — Updated daily*