Usage Examples & API

Query the daemon’s live state

curl http://localhost:8788/state

Returns the full DeskSnapshot: status, active fixture, per-fixture decisions (last 200), quotes, radar events, anchor evidence, grade sheet, and portfolio risk.

Verify the hash chain

curl http://localhost:8788/verify

{ "ok": true, "fixtures": [{ "fixtureId": "18209181", "ok": true, "headHash": "…" }] }

Request a Merkle inclusion proof for one decision

curl "http://localhost:8788/ledger/proof?fixtureId=18209181&seq=47"

See Verifiability for how to verify the response locally.

Run the Strategy Arena on demand

curl "http://localhost:8788/arena?fixtureId=18209181"
curl "http://localhost:8788/arena/ablation?fixtureId=18209181"

Both run on demand over the fixture’s authoritative captured corpus — not a second continuously running live session.

Get a shareable grade card

curl "http://localhost:8788/grade-card.svg?fixtureId=18209181" -o grade-card.svg

Check what policy was live, and when it changed

curl http://localhost:8788/policy/snapshots

Run calibration/holdout evaluation

pnpm --filter @tissue/daemon evaluate:calibration

Get human-reviewed tuning suggestions

pnpm --filter @tissue/daemon evaluate:tuning

Never writes policy.toml — see Autonomous Operation.

Query the analyst layer

The analyst exposes 8 read-only MCP tools. Example: find structurally similar past decisions to a reference decision.

// via MCP client, tool: find_similar_decisions
{
  "fixture_id": "18209181",
  "seq": 47,
  "minute_tolerance": 10,
  "edge_tolerance_bps": 100
}

Every row returned is a real, citable ledger record — matched on radar class, match-minute proximity, and edge-magnitude proximity, not embeddings.

Dashboard walkthrough

Page	Shows
`/overview`	Desk status, live streams, portfolio risk
`/decisions`	Decision feed with regime badges, hash-chain verify button, on-chain commitment timeline, TxLINE input verification list, real Slip execution evidence
`/radar`	Signal-class legend and live latency-radar events
`/quotes`	Quote tape with receipt links (decision hash, proof messageId, explorer URL)
`/grade`	Grade sheet, shareable grade card, equity curve
`/arena`	Strategy Arena head-to-head, regime ablation matrix
`/verify`	In-browser third-party verifier — recomputes a decision’s hash chain and fetches the anchoring transaction directly from a public Solana RPC
`/analyst`	Read-only LLM forensics chat over the ledger

Execution disclaimer

Within TxLINE’s own on-chain program, live mode publishes risk-approved quotes only — txoracle has no intent-book to fill against, so it never invents a fill, a counterparty, or PnL there. Simulated fills exist only in replay/research mode and are always marked simulated: true, end to end, in the ledger, the API, and the dashboard. Real execution does exist on a separate venue (Slip), off by default and gated by its own stricter risk check — see Verifiability.

Get Started

How It Works

Build & Run

Reference

Usage Examples & API

Query the daemon’s live state

Verify the hash chain

Request a Merkle inclusion proof for one decision

Run the Strategy Arena on demand

Get a shareable grade card

Check what policy was live, and when it changed

Run calibration/holdout evaluation

Get human-reviewed tuning suggestions

Query the analyst layer

Dashboard walkthrough

Execution disclaimer