Why Conyr for Agents

Conyr gives an AI agent computed on-chain intelligence as callable tools — not raw transactions to crunch. Ask “is this wallet skilled or just lucky?”, “who’s coordinating this token?”, “is this launch farm-seeded?” and get a structured, evidence-linked answer in one call. Every tool is the same data and tiers as the REST API, wrapped for agents.

Connect in one of three ways

MCP Server (fastest)

~46 tools over the Model Context Protocol. One command, no glue code.

OpenAPI & llms.txt

Generate a typed client, or feed the spec/llms.txt to your model as context.

Tool Catalog

Every tool, what it answers, and the tier it needs.

Built for agents

Structured + summarized + cited. Every tool returns structured_content (full JSON for programmatic use), a one-line text summary (for plain-text models), and an evidenceUrl pointing at the Conyr terminal page that backs the answer — so an agent can always cite its source.
One-call dossiers. Composite tools (token_deep_dive, wallet_overview, token_coordination_report, wallet_network_report) fan out many readers and return a complete dossier in a single call — fewer round-trips, fewer tokens spent on orchestration. Reach for these on broad questions; use the targeted query_* tools for follow-ups.
Signature answers other APIs can’t give. Position-correct trader skill (query_wallet_quality — skilled vs. lucky), real-time bundle/coordination detection, funding-abuse diagnostics, and provenance.
Graceful degradation. A single store hiccup degrades one section of a dossier to null (with an errors map) rather than failing the whole call.

⚠️

Detection, not trading signals. The coordination, minority-buyer, and forming-push tools describe what is happening on-chain — coordinated buying, farm clustering, wash setups. The same patterns describe manipulation and organic accumulation; Conyr makes no claim about which, and none of it is financial advice. Answer in a neutral, observer voice.

What the tools cover

Domain	What an agent gets
Wallet skill & PnL	Position-correct skill score, realized PnL, performance, positions, trades
Wallet identity	Behavioral classification + badges, behavioral-similarity look-alikes, funding provenance
Coordination	Follow edges, funding/execution entities, campaigns, crowd state, entanglement
Token risk	Security verdict, health score, bundles, wash trades, funding abuse, buyer quality
Holders & market	Holder concentration + label breakdown, OHLCV, trade history

The underlying field models live in the API Reference. DEX and router coverage spans every major Solana venue — see Data Model & Enums.

The basics

Base URL   https://api.conyr.ai
Auth       Authorization: Bearer <API_KEY>   (all /v1 routes)
Responses  application/json · errors: {"error": "...", "code": <int>}
Pagination limit + offset (no cursors)

Machine-readable resources

OpenAPI 3.1

The full REST contract — generate a client.

llms.txt

Compact context for an LLM.

llms-full.txt

Full context dump.

💡

Winner, Infrastructure Track — Colosseum Breakout Hackathon.

MCP Server