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

# Technical Reference

> TxLINE endpoints, data schemas, configuration, and on-chain program details.

## TxLINE endpoints used

| Endpoint                                                                 | Purpose                                                                         |
| ------------------------------------------------------------------------ | ------------------------------------------------------------------------------- |
| `POST /auth/guest/start`                                                 | Expiring guest JWT + renewal                                                    |
| `POST /api/token/activate`                                               | Activated API token (handles both observed plain-text and JSON response shapes) |
| `GET /api/scores/stream`                                                 | Live match state, SSE                                                           |
| `GET /api/odds/stream`                                                   | Live StablePrice consensus odds, SSE                                            |
| `GET /api/scores/snapshot/{fixtureId}`                                   | Historical/activation capture support                                           |
| `GET /api/odds/snapshot/{fixtureId}?asOf={ts}`                           | Historical odds capture                                                         |
| `GET /api/odds/validation?messageId={id}&ts={ts}`                        | Merkle proof for `validate_odds`                                                |
| `GET /api/scores/stat-validation?fixtureId={id}&seq={seq}&statKey={key}` | Score-stat Merkle proof for `validate_stat`                                     |

## On-chain program IDs

| Network | TxLINE oracle program                          |
| ------- | ---------------------------------------------- |
| Devnet  | `6pW64gN1s2uqjHkn1unFeEjAwJkPGHoppGvS715wyP2J` |
| Mainnet | `9ExbZjAapQww1vfcisDmrngPinHTEfpjYRWMunJgcKaA` |

| Anchoring                                   | Program                                                  |
| ------------------------------------------- | -------------------------------------------------------- |
| Pre-Match Commitment + periodic checkpoints | SPL Memo — `MemoSq4gqABAXKb96qnH8TysNcWxMyWCqXgDLGmfcHr` |

## Daemon API endpoints

| Endpoint            | Method | Purpose                                                 |
| ------------------- | ------ | ------------------------------------------------------- |
| `/health`           | GET    | Liveness                                                |
| `/ready`            | GET    | Readiness (verified proofs present, no recent failures) |
| `/state`            | GET    | Full desk snapshot                                      |
| `/verify`           | GET    | Hash-chain verification per fixture                     |
| `/arena`            | GET    | Strategy Arena head-to-head                             |
| `/arena/ablation`   | GET    | Regime ablation matrix                                  |
| `/grade-card.svg`   | GET    | Shareable grade card                                    |
| `/ledger/proof`     | GET    | Merkle inclusion proof for one decision                 |
| `/policy/snapshots` | GET    | Signed policy change log                                |
| `/metrics`          | GET    | Prometheus-format metrics                               |
| `/events`           | GET    | Read-only SSE evidence stream                           |

## Analyst MCP tools

| Skill                      | Tools                                                                                                 |
| -------------------------- | ----------------------------------------------------------------------------------------------------- |
| `ledger-forensics`         | `get_recent_decisions`, `get_signal_class_stats`, `query_ledger_by_fixture`, `find_similar_decisions` |
| `slip-market-intelligence` | `list_slip_markets`, `inspect_slip_market`, `list_slip_wallet_tickets`                                |
| `slip-settlement-audit`    | `verify_slip_market_reference`, `inspect_slip_market`                                                 |

### `find_similar_decisions`

Structured pattern recall over the read-only ledger projection: same radar class (or
same action, if the reference has none), match-minute within tolerance, and edge
magnitude within tolerance, ranked by combined distance. Deliberately not an embeddings
claim — see [Autonomous Operation](/autonomous-operation).

## Decision record schema (summary)

```typescript theme={null}
interface DecisionRecord {
  seq: number;
  triggerMsgId: string;
  triggerHash: string;
  triggerNetwork: "devnet" | "mainnet";
  ts: number;
  action: "POST" | "REPLACE" | "CANCEL" | "NO_ACTION" | "HALT";
  radarClass?: string;
  haltReason?: string;
  state: StateSnapshot;      // minute, score, cards, inventory, exposure, regime flags
  tissueProb: number;        // bps
  marketProb: number;        // bps
  edgeBps: number;
  intents: Intent[];
  anchorTxSig?: string;
  simulated: boolean;
  prevHash: string;
  hash: string;
  signature?: string;        // Ed25519, hex
  signerPubkey?: string;     // base58
}
```

## Key configuration (`policy.toml`)

Every tunable constant in the pricing and risk logic lives here — no magic numbers
outside it. Selected fields:

```toml theme={null}
[strategy]
edge_threshold_bps = 150          # minimum |tissue_p - market_p| to quote
base_spread_bps = 80
stoppage_spread_mult = 1.3
mutual_danger_spread_mult = 1.5

[risk]
drawdown_kill_units = 3_000_000_000
portfolio_drawdown_kill_units = 7_000_000_000
proof_failure_window = 20
proof_failure_min_samples = 8
proof_failure_rate_halt = 0.5

[exec]
checkpoint_interval_decisions = 20   # periodic on-chain checkpoint cadence
anchor_sample_rate = 1.0
```

Config validation is an exhaustive recursive shape-check against every field — a
missing field fails loudly at boot instead of silently disabling a risk gate.