Product Architecture

Built for Bring-Your-Own Agents.

OutreachAgent handles email state machines, deliverability, webhooks, and thread retrieval so your agent runtime can focus on reasoning and decision-making.

Optional Reply Enrichment

If you enable intelligence helpers, OutreachAgent can extract structured metadata from inbound content before it reaches your agent. This is optional enrichment, not core runtime logic.

Thread History Retrieval

Conversation history stays queryable and durable so your runtime can fetch the exact context it needs at execution time.

Dynamic Rate Limiting

Agents can go rogue. We enforce hard sender limits per domain and per IP so a reasoning error doesn't nuke your reputation score.

Automated IP Pacing

Connect 50 domains and we handle the warming schedules. Our system naturally ramps up volume day by day, protecting your deliverability.

Spam Trap Avoidance

We automatically cross-reference outbound targets with known spam trap databases and bounce lists before the message is sent.

Native Agent Tools

Use typed SDKs, an MCP server for Cursor and Claude Desktop, or direct API calls. Keep your agent runtime outside the platform — OutreachAgent handles the mail primitives.

Portable agent skills

Native tools for execution, public skills for judgment and voice.

OutreachAgent handles inboxes, workflows, and delivery state. Portable skills like `cold-outreach-writer` handle research, voice control, subject-line discipline, and sequence planning without locking users into one runtime.

Install command

npx skills add https://github.com/pagefarms/cold-outreach-writer.git --skill cold-outreach-writer

Deep Integration with OpenClaw

If you use OpenClaw, we provide a tool wrapper. If you do not, use the API directly. Either way, the runtime stays yours and OutreachAgent stays the infrastructure layer.

import { OutreachAgentTool } from '@outreachagent/openclaw';

const agent = new OpenClawAgent({
  role: 'SDR',
  tools: [OutreachAgentTool.getCapabilities()],
});