Cookbooks
Step-by-step recipes that take you from nothing to using an AgentBeam feature fully, with commands to copy and a prompt to hand the whole recipe to your coding agent.
On this page
The guides under Agent CLI explain what each feature does. Cookbooks are the other half: an ordered recipe for one job, from a blank setup to every part of the feature switched on and checked, with commands you can paste as they are.
Hand a cookbook to your coding agent
Every page on this site is also plain Markdown at the same address plus .md, for example docs.agentbeam.com/cookbooks/app-monitoring.md. Copy for agents, at the top of each page, copies a prompt that points Claude Code, Codex, Cursor or any agent that can fetch a URL at that Markdown and asks it to work through the recipe with you. You can also paste the prompt yourself:
Read the AgentBeam cookbook at https://docs.agentbeam.com/cookbooks/app-monitoring.md and work through it with me step by step for this project: run the commands, make the code changes it describes, and confirm each check passes before moving on. Ask me before anything that needs a secret or an admin.
Agents that read llms.txt find every guide and cookbook there, each linked to its Markdown.
Recipes
| Cookbook | You end with |
|---|---|
| Agent monitoring | Every coding agent on every developer machine hooked and captured, MCP servers inventoried, a policy moved from observe to enforce, and a weekly review |
| App monitoring | A running app watched for attacks, with security events, LLM checks, dependency checks in CI, alerts that reach the right people, and a routine for reading what comes in |
| Agent evaluation | Each agent action judged by a model or service you choose, tested with sample actions in CI, and the agent stack benchmarked with WipeBench |
| Skill testing | Skills, MCP configs and packages scanned before agents use them, risky skills gated at load time, an inventory, and a CI gate |
| Skill sharing | A reviewed team library of skills, instructions, commands and subagents that every teammate installs and keeps current |
By framework
The general recipe covers every App monitoring feature. These show the same setup in each framework's own terms: where to hook requests so their status is known, which auth hooks report sign-ins, where the logs go, and which framework rules Beam turns on.
| Framework | How Beam sees it |
|---|---|
| Next.js | SDK in proxy.ts, Auth.js events, LLM route checks |
| Express and Fastify | SDK after each response |
| NestJS | SDK in a global middleware |
| Django | Request middleware and auth signals, read by beam monitor |
| Flask | after_request hook and login view, read by beam monitor |
| FastAPI | HTTP middleware and auth endpoint, read by beam monitor |
| Ruby on Rails | Rack middleware and Warden hooks, read by beam monitor |
| Laravel | Terminable middleware and auth events, read by beam monitor |
| WordPress | Web server access log and a must-use plugin |
| Spring Boot | Tomcat access log and Spring Security events |
| Go | net/http middleware with slog |
| ASP.NET Core | Middleware with Serilog and Identity sign-ins |
More recipes follow the same shape: what you'll have at the end, what you need first, numbered steps with a check after each, and a coverage checklist at the end so nothing is left switched off.
How a recipe is laid out
- What you'll have lists the end state, so you can tell when you're done.
- Before you start names the roles and access you need. Steps that need a workspace owner or admin say so.
- Steps are in order. Each ends with a check: a command whose output, or a page in the dashboard, shows the step worked.
- Coverage checklist is every feature the recipe touches, in one table. Tick it off, or hand it to your agent to verify.
Secrets never go in a command line or a repository. Where a recipe needs a token or key, it reads it from your platform's secret store or an environment variable.
Set this up with your coding agent
Install the agentbeam-setup skill for Claude Code, Codex, Cursor, Gemini CLI, OpenCode and Copilot, then ask your agent to set up AgentBeam on this machine. Beam scans the skill before it installs.
npx -y @agent-beam/beam add skill agentbeam-setup -g --all-agents