Skip to content
Docs menu

Cookbook

End-to-end recipes that put venom's pieces together, from an offline coding setup to a multi-tenant SaaS with a phone agent.

View .md

Each recipe here starts from nothing and ends with something running. The steps are copy-pasteable, and every flag, config key, environment variable and export they use exists in this release. A recipe links to the reference page for each piece instead of repeating it, so read the linked page when you need the options a recipe leaves out.

Recipe You end with Pieces
A collections SaaS on venom a multi-tenant back office whose agent proposes and a person approves, per-tenant keys, budgets and guardrails, an evidence trail, and the plan for a phone agent harness, orchestrator, guardrails, inbox, voice
Embed the harness in a Next.js app a Next.js 16 app that streams an agent's answer to the browser, then the same app talking to a venom serve worker with approvals in the UI harness, harness server
Run venom fully offline with Ollama a coding agent on local models with nothing leaving the machine, and how to prove it CLI, Ollama, orchestrator, air-gapped bundle
Venom Cloud for a team one endpoint for the team, a key and budget per person, your own GPU box joined as a daemon Venom Cloud, CLI, orchestrator in mesh mode
Use venom from another agent venom's browser tools, its web search and your own venom agent as MCP servers for Claude Code, Codex or Cursor venom browser mcp, the orchestrator's /mcp, the harness's MCP server
Add a provider key and route by cost a cloud provider next to your local models, an alias that tries the cheapest target first, and spend you can read back orchestrator upstreams, routes, pricing, profiles, gears

Before you start

Most recipes need the venom CLI (install) and a model endpoint (local or cloud). The builder recipes also need the harness libraries, which are tarballs rather than registry packages (install the harness).

Commands that need a source checkout of venom (make …) say so; everything else works from a release install.

In this section

  • Build a collections SaaS on venom: Put the harness, the orchestrator, guardrails, human handoff and the compliance evidence together into a multi-tenant collections product, with the phone agent's design and what of it ships today.
  • Embed the harness in a Next.js app: Build a Next.js 16 app that streams a venom agent's answer to the browser, then move the sessions to a venom serve worker and answer the agent's approvals from the page.
  • Run venom fully offline with Ollama: Run the venom coding agent on local models with no network, from Ollama alone up to the orchestrator and an air-gapped install, and check that nothing left the machine.
  • Put venom behind Venom Cloud for a team: Give a team one Venom Cloud endpoint, a key and budget per person, provider keys or your own GPU box as a daemon, and the shared policy, then point everyone's venom at it.
  • Use venom from another agent over MCP: Give Claude Code, Codex, Cursor or your own agent venom's browser tools, its web search and a whole venom agent as MCP tools, and know which other interfaces fit better.
  • Add a provider key and route by cost: Add a cloud provider to your orchestrator next to local models, make an alias that tries the cheapest target first, choose it per key, request or gear, and read back what it cost.