Blockchain intelligence should be usable by agents and auditable by humans
Chain Insights turns semantic blockchain data into investigation infrastructure: Chain Insights Graph for context, MCP for user-run agents, and ACP for agent-to-agent commerce on Virtuals Protocol.
Topology, facts, risk scores, labels, and investigation context.
Self-run investigations for Codex, Claude Code, Hermes, and custom agents.
Virtuals Protocol agent-to-agent commerce for Chain Insights workflows.
Investigation capabilities without the table wall
The public product surface is simple: screen risk, trace funds, inspect graph context, and prototype locally before using hosted access.
graph_queryUse the base read-only Graph surface for live topology, archive topology, and facts. The cia package builds named investigation tools and reusable queries on top of it.
Screen an address for risk context, behavior, neighborhood signals, and exchange exposure.
Trace backward from a suspected deposit or cashout point to upstream sources, shared funders, and convergence.
Trace suspected scammer, mule, operator, or laundering funds forward toward cashout topology.
Trace victim or trusted-source funds through intermediaries toward exchange deposit candidates.
Run a Chain Insights-compatible backend with bundled Bittensor data for local agent and workflow experiments.
Depth over breadth
Chain Insights runs one network end to end — Bittensor today. Every block is indexed, every address resolved to an identity, and every flow enriched and risk-scored on a single pipeline, so every trace, screen, and query reads the same full-history source of truth.
Substrate and EVM indexers ingest every block, event, and transfer from the network into a complete, queryable history.
Data pipelines transform raw history into a semantic model: addresses clustered into identities, transfers aggregated into flows, and known actors labeled.
The enriched topology, facts, and evidence become the Chain Insights Graph — queryable across full history and a live view of the network.
Machine-learning models propagate risk from confirmed labels across the graph and score every identity and its neighborhood.
Trace, screen, query, and graph tools are exposed over MCP for any capable agent — Codex, Claude Code, Hermes, or your own — to call.
The cia CLI runs investigations locally and preserves every result — graph JSON, tables, and case notes — as durable, replayable evidence.
Graph, MCP, and ACP are access paths
This is not a SaaS pricing table. Start with graph access, run your own MCP-connected workspace, or let agents transact through ACP.
Use Chain Insights Graph for topology, facts, risk scores, and custom graph questions.
Connect Codex, Claude Code, Hermes, or your own MCP-capable agent.
Virtuals Protocol agent-to-agent commerce for Chain Insights investigation workflows.
Live now. Expanding continuously.
The public surface stays simple: what works today, what is in progress, and what continues to grow.
Live topology, facts, and risk scoring.
Semantic graph context for investigations.
User-run agents can trace, screen, query, and preserve evidence.
Agent-to-agent commerce for Chain Insights workflows.
Backing the mission
$CIA aligns the product, team, and community around Chain Insights as open agentic AML infrastructure, with product revenue supporting buybacks.