oper8r Docs
Customer-facing implementation guide
oper8r exposes organization-approved tools through MCP so LLM clients can search, reason over account context, explore the graph, run GTM workflows, and, when explicitly enabled, curate the organization-internal employee graph.
MCP is useful when your team already works in ChatGPT, Claude, Claude Code, Codex, Cursor, or another agent client and wants that client to use oper8r context safely.
Use the organization-scoped MCP URL when configuring a client:
https://YOUR_OPER8R_HOST/api/mcp/org/YOUR_ORG_SLUG
For hosted oper8r, this is usually:
https://oper8r.io/api/mcp/org/YOUR_ORG_SLUG
Generic MCP client configuration:
{
"mcpServers": {
"oper8r": {
"url": "https://YOUR_OPER8R_HOST/api/mcp/org/YOUR_ORG_SLUG"
}
}
}
Some clients will prompt for OAuth or browser-based login. Others may require an approved token or connector configuration depending on deployment.
MCP tools can be enabled per organization and role.
Common tool groups:
See MCP Tools for the current tool categories.
MCP tool definitions tell a client what tools exist. Agent skills tell the client how to use those tools well.
Install oper8r skills with:
npx @oper8r-io/agent-skills install all --client all
For skill-specific setup, see Agent Skills.
oper8r tools are not automatically exposed to every user. Admins can grant tools for an organization and set minimum roles. A user only sees the tools that are both enabled for the organization and allowed for their role.
Mutating tools should be granted more carefully than read and synthesis tools. Employee graph curation tools are explicitly scoped to the shared internal employee graph and should not be used for private user memory or customer-facing public graph data.
Use oper8r to prepare a renewal brief for Acme.
Include recent call themes, CRM stage and owner context, open risks,
stakeholder coverage, and the best cited next actions.
Bring us the workflow, integration, or deployment constraint. We can build it with you, run it for you, or advise your team.