Skill

Use this section when you want AgentQL to be consumed as an installable host skill instead of a raw transport surface.

The public install artifact lives in AgentQL/skills, but this docs section is the current usage guide. It explains how to use the skill with the same route truth as the rest of this docs site.

Install artifact

• Repository: AgentQL/skills
• Current install command: npx skills add AgentQL/skills -g -y
• API key source: agentql.tomo.inc
• Current endpoint source: agentql.tomo.inc/nodes

Read by host

• IDE Agents: read the Codex, Cursor, and generic IDE-host workflow in one grouped section instead of mixing host pages into the parent menu.

Read by agent role

• Install & Configure: install the skill, set the API key, and keep endpoint truth externalized.
• Market Research Agent: use the skill for latest price, market profile, asset resolution, and OHLCV reads.
• Monitoring Agent: use the skill for continuous market checks, watches, and live event workflows.
• Chain Data Agent: use the skill when the runtime needs native EVM JSON-RPC methods or subscriptions.
• MCP Agent: use the skill when the host can already work with MCP-like tool workflows and you want clearer prompt routing.
• Prompt Patterns: copy prompt shapes that keep the skill inside the current public contract.

When to use Skill instead of SDK or raw transports

• Use Skill when the host already has a skill or tool-install model and you want the shortest host-level setup.
• Use SDK when your application should call AgentQL directly from a typed TypeScript runtime.
• Use Getting Started when you still need to compare raw HTTP, MCP, Market WebSocket, and EVM RPC directly.

Boundary

• The skill should route users toward hosted product surfaces, not direct upstream providers.
• The endpoint source of truth stays on agentql.tomo.inc/nodes.
• Current public docs lead with EVM-family routing. Solana remains Beta.