What is live, what is being wired, what is next.

Describe what you want to know about a launch, creator or pool. No query syntax.

The objective is tokenised, classified by intent and routed onto a decomposition pattern. The parse is shown to you rather than hidden, so the routing decision is inspectable before anything runs.

Every objective becomes three to five concrete subtasks with their own read surfaces.

Decomposition is the unit of assignment. Each subtask names the chain surfaces it needs, which is what turns agent recommendation into a routing decision rather than a label match.

A recommended team you can change, with the trade-off made explicit.

Team size is a real decision. One agent is faster and narrower; three cover more surfaces and corroborate more claims. The brief states which was chosen and adjusts its confidence accordingly.

Selected agents are built from their compartments before deployment.

The manufacturing sequence is the moment an abstract team becomes a concrete set of units with assigned subtasks and named read surfaces.

One organised result rather than several disconnected outputs.

Findings, patterns, contributions, coverage and a relationship graph are reconciled into a single document with per-claim confidence and per-agent attribution.

TokenLaunched events resolved to token, creator, pool and self-described metadata.

The pons factory emits a launch event carrying everything needed to start a mission. Indexing it continuously is what makes a launch visible at block one instead of when a feed refreshes.

Uniswap V3 pool configuration, slot0 pricing, paired ETH and locked position.

Launches pair into a WETH pool at the 1% tier with the position held by the locker. Reading it directly is the difference between knowing a price and knowing how much depth stands behind it.

Swap events with direction, size, wallet and block ordering retained.

Volume is the metric that hides structure. Keeping direction, size and wallet on every swap is what lets an agent tell a broad pool from a narrow one.

Fixed supply, lock state and pool configuration re-read before a claim is retained.

A claim that cannot be matched to contract state is downgraded and the downgrade is recorded. Nothing is retained on the strength of a launch event alone.

A creator address across its launches, fee accrual and claim behaviour.

Creators are the most durable entity on a launchpad. Indexing them turns a single token page into a track record.

Holder distribution derived from transfers, and overlap measured between launches.

Holder overlap between a creator's launches is the strongest structural signal on the chain and the one least visible from any single token page.

The socials a token publishes on its own contract, resolved and weighted.

Tokens self-describe their channels, which makes the links trivially readable and trivially empty. Resolution separates a real venue from a populated field.

Trading venues on Robinhood Chain carrying the pair, and how depth is distributed.

Robinhood Chain has a growing venue set. Knowing where else a pair trades is what stops a single-pool read from being mistaken for the whole market.

Agents read one another's intermediate findings during a mission.

An append-only mission context lets a later agent extend an earlier agent's work instead of repeating it, and makes disagreement detectable at all.

Explicit transfer of a subtask between specialised agents.

Handoffs carry the reason for the transfer, which is what makes a multi-agent trace readable after the fact.

Watch a running mission and redirect it before it finishes.

Pausing, redirecting and cancelling a mission in flight, with the partial result preserved rather than discarded.

Re-run a mission on a cadence and compare against the last run.

Monitoring is a mission with a cadence and a diff. On a launchpad the interesting output is what changed since the last block range, not what was found again.

Alerts on graduation distance, wallet breadth or holder concentration crossing a stated line.

Thresholds are defined against the mission's own measurements, so an alert can always be traced back to the finding that produced it.

Shared missions, shared monitors and per-member permissions.

Missions become organisational assets rather than personal ones, with an audit trail of who deployed what against which block range.

Programmatic mission creation and structured result retrieval.

The same engine contract the interface uses, exposed as an API. Mission in, structured brief out.

Define an archetype with your own capabilities and read surfaces.

An archetype is a role, a capability list and a source scope. Making that authorable is what turns the product into a platform.

The same mission model pointed at launch venues beyond pons.

The archetypes are not pons-specific in shape, only in scope. Extending to another venue is a connector problem, not an interface problem.