Loading
Loading
Loading
Loading
Loading
Loading
Loading
Pricing and data
What this deployment computes, what it only describes, and where each number came from.
What the engine returns, and what it refuses
No list
The engine’s refusals were not read, so this page prints no count. A zero here would be a measurement nobody made.
This build reads its bundled fixtures, so there is no refusal list to read. (no_backend)
What was run
Two recorded runs, kept apart. The first ran against the v2 program on a local validator; the second is the earlier dry run, simulated inside itself and labelled as such.
SYNTHETIClive-onchain-swarmswarm-summary
6 agents on scripted policies (1 buyer, 1 greater fool, 1 holder, 1 panic, 1 scalper, 1 sniper), 10 of 10 configured rounds over 2 events, on the local validator this machine started, against program v2.
The program is v2 at CTEnej2b5cgJ7uQ185k7yWZChe8GasAgPYCc4BHFB67w, the build that local validator ran. It is not the program the page at /rules prints, which is the devnet deployment 5EJ1GU3M…4eva. Two programs on two chains, and nothing on this panel is evidence about that one or about any public cluster.
- run
- record
- seed
- agents
- chain
- cost
- run wall_seconds
- budget
the aggregate the run's summary tool writes from its two datasets; the datasets stay out of the repository
1 buyer, 1 greater fool, 1 holder, 1 panic, 1 scalper, 1 sniper
cluster local · dry_run false
scripted policies: no model calls, and no cluster spend -- the validator is funded at genesis and nothing is rented
the record's own value, the same as the second event's wall; the record carries no separate total
spend summed over the events against one stated pot: each sidecar's counter is that event's, so the events are added and the pot is not
4 refused on chain: 6010 E_CAP_EXCEEDED × 4
0 declined by policy
10 bids no ask met and 0 asks no bid met, kept apart because they are opposite sides of the book. The executed count is trades only: a listing is a price an agent asked for and no unit moved.
capped
evt_cappedcap 2500 bps · ceiling 1,250,000,000 lamports · face 1,000,000,000 lamports
10 of 10 rounds completed · 28 rows · 10,600,000,000 lamports notional
- refused on chain
- refused by code
- declined by policy
- by action
- executed by agent
- refused by agent
- unfilled
- wall_seconds
buy, list and resell as the record counts them; a listing is not a trade
The vertical scale runs from zero to 1,500,000,000 lamports, the highest price this event’s record holds. The line is the cleared median, the ticks are the highest price asked in each round, the dashed rule is the ceiling, and a round with no trade is marked n 0 with no line through it.
cleared. lower median (nearest rank) price that transacted in the round: trades only (buy, resell, status ok). A round in which only listings and refusals happened carries this point with median_lamports null, max_lamports null and n 0, so a chart draws a gap rather than a line through a round that never traded
asked. lower median (nearest rank) and highest price an agent named in the round -- bids, resale asks and listings, refusals included. A listing is a price an agent asked for with no chain instruction behind it, so it appears here and never in executed or cleared
monitored
evt_monitoredcap none (65535 is the harness's uncapped sentinel) · ceiling 7,553,500,000 lamports · face 1,000,000,000 lamports
10 of 10 rounds completed · 28 rows · 15,200,000,000 lamports notional
- refused on chain
- refused by code
- declined by policy
- by action
- executed by agent
- refused by agent
- unfilled
- wall_seconds
buy, list and resell as the record counts them; a listing is not a trade
The vertical scale runs from zero to 7,553,500,000 lamports, the highest price this event’s record holds. The line is the cleared median, the ticks are the highest price asked in each round, the dashed rule is the ceiling, and a round with no trade is marked n 0 with no line through it. The scale is set by the ceiling rather than by a price, so the recorded prices sit low in the frame: nothing in this event came near the ceiling.
cleared. lower median (nearest rank) price that transacted in the round: trades only (buy, resell, status ok). A round in which only listings and refusals happened carries this point with median_lamports null, max_lamports null and n 0, so a chart draws a gap rather than a line through a round that never traded
asked. lower median (nearest rank) and highest price an agent named in the round -- bids, resale asks and listings, refusals included. A listing is a price an agent asked for with no chain instruction behind it, so it appears here and never in executed or cleared
The record’s own limits
Synthetic data only: scripted agents trading a synthetic ticket on a local validator. No real market price, markup or refusal rate is measured here, and the chain the rows name is a validator this machine started, not a public cluster.
Generated by agents against a local validator or in simulation. This is NOT evidence of real market pricing and must not be used to estimate real prices.
Every figure here was read from records/swarm-live-onchain.json at build time by tools/live_run_panel.py, which writes frontend/data/live-onchain-run.json and regenerates it on demand. This page is a static export: nothing was re-read when a browser opened it, and no validator was asked while it was built.
The earlier scripted dry run: 9 agents, 12 rounds, dry_run
SCRIPTED AGENTS · DRY RUN · SYNTHETICIts two charts belong to that run and to no other, and no number in it is a number from the panel above. Its own chain field says this: dry run: the chain is simulated in the run, not a validator.
- run
- recorded
- agents
- cost
2 buyer, 2 flipper, 1 greater fool, 1 holder, 1 panic, 1 scalper, 1 sniper
scripted agents, no model calls, so no dollars were spent
capped
cap 2500 bps · ceiling 1,250,000,000 lamports51 decisions over 12 rounds. 32 tickets.
monitored
cap none (65535 is the harness's uncapped sentinel) · ceiling 7,553,500,000 lamports106 decisions over 12 rounds. 32 tickets.
Transcribed by tools/swarm_panel.py from the run's own meta files, which are not committed because they name this machine's scratch directory. The two charts are byte copies of docs/charts/price-path-*.svg from the same run.
Computed, described, and not built
- available
- kind
- fitted
- modelVersion
No model is registered.
Not available
Mock mode reads no pricing engine, so there is no model to describe.
Computed
A price bound from this event’s own resale record, and the cap comparison in whole lamports. Both are arithmetic, and both are reproducible from the rows on screen.
Described
The demand response and the fee model. The engine refuses both rather than fitting either, so nothing here is a number it made up to fill the shape.
Not built
A sale probability, an upper bound, and an asking price: no listing was ever recorded, so no row anywhere holds what a seller wanted.
Nothing was fitted behind the price bound, and the bound itself is unvalidated. The refusal list above is what makes that checkable line by line.
Where the data comes from
- Mode
- Synthetic
dataSource BUNDLED MOCK DATA. Generated from lib/mock on this machine, not read from the backend.
The chain
What this build was told, what it asked, and what a verified artefact records. These are three different things and they are printed as three different fields.
- programDeployed
- rpcEnabled
- askedCluster
- cluster
- reason
- detail
- No backend is configured, so this process asked no cluster and read no artefact. Nothing here is a deployment claim.
This build queried no cluster, so programDeployed is the absence of a check rather than the result of one.
The contract
How much of the documented contract the build’s own health payload says it implements.
- endpointsInContract
- endpointsImplemented
- routesImplemented
Counted against API_CONTRACT.md.
The guidance console
The console is served by the backend at its own origin. No backend origin is configured in this build, so there is nothing here to link to.