Build with Coding Agents

Use Claude Code, Codex, or Cursor to set up, build, improve, evaluate, review, and deploy an AgentOS Starter. The repository gives your coding agent the source code, a live MCP endpoint, evals, and runtime logs in one place. It can make a change, exercise it against the running platform, and bring back a verified result. Everything on this page runs inside a cloned AgentOS Starter. If you don’t have one running yet, pick a Starter first.

Skills

Each AgentOS Starter stores its coding-agent workflows in .agents/skills/.

Skill	Use it when
`/setup-platform`	Bring up a fresh clone, verify the platform, and build the first agent
`/create-agent`	Turn an idea into a registered agent and smoke-test it live
`/extend-agent`	Add a tool, capability, instruction change, or targeted fix
`/improve-agent`	Derive probes from an agent’s instructions and harden its behavior
`/create-evals`	Add eval coverage for an agent from its behavior and real sessions
`/eval-and-improve`	Diagnose failing evals and repair the affected behavior
`/review-and-improve`	Check code, documentation, configuration, and registered components for drift
`/deploy-platform`	Deploy with the template’s provider workflow and verify the live platform

The setup prompt invokes the setup-platform skill from .agents/skills/. Claude Code also discovers the skills through the Starter’s committed .claude/skills symlink. Codex and Cursor can link to the same .agents/skills/ directory.

Connect the live platform

Connect the local AgentOS to the supported clients on your machine:

uvx agno connect

The command registers AgentOS with detected Claude Code, Claude Desktop, Codex, and Cursor clients. Your coding agent can then call run_agent, run_team, and run_workflow while it works on the repository. For a deployed platform:

uvx agno connect --url https://<your-domain>

See Connect Your Clients for authentication modes and client-specific setup.

Keep behavior stable with evals

Use the bundled eval suite to check response quality and tool calls as you change instructions, tools, and models. The evals run on the host machine, so set up the virtual environment once:

./scripts/venv_setup.sh
source .venv/bin/activate

Then run:

python -m evals --tag smoke      # fast checks of the platform agents
python -m evals --tag release    # broader pre-release confidence
python -m evals --name <case>    # one case while iterating
python -m evals -v               # stream the full run with rich panels

Cases wrap AgentAsJudgeEval (LLM judge, binary pass/fail) and ReliabilityEval (tool-call assertions). Results log to Postgres, so run history shows up at os.agno.com next to sessions and traces. If a case fails, run /eval-and-improve.

Agent patterns

The bundled agents demonstrate three patterns to copy:

Pattern	Example	When to use
Direct tools	`agents/web_search.py`	The agent needs fine-grained control over each tool call.
Context provider	`agents/platform_manager.py`	You want many tools collapsed into one `query_<thing>` interface that hands off to a sub-agent.
Studio builder	`agents/agent_builder.py`	Users create and refine components from chat; deletes keep a confirmation gate.

Scheduled tasks

The scheduler is on by default. Two reference workflows are registered out of the box:

Workflow	What it does	Default
Deployment check	Daily readiness report: database, auth, scheduler, MCP, Slack config. Fixed checks, no LLM calls.	On. Disable with `ENABLE_DEPLOY_CHECK=False`.
Run evals	Daily smoke-tag eval run. Uses model calls.	Off. Enable it from the AgentOS UI.

See Scheduler for the cron API.

Next steps

Task	Guide
Choose a Starter	AgentOS Templates
Connect coding clients	Connect Your Clients
Add or customize evals	Evals

Get Started

Starters

Apps

Agno CLI

Build with Coding Agents

Skills

Connect the live platform

Keep behavior stable with evals

Agent patterns

Scheduled tasks

Next steps

​Skills

​Connect the live platform

​Keep behavior stable with evals

​Agent patterns

​Scheduled tasks

​Next steps

Skills

Connect the live platform

Keep behavior stable with evals

Agent patterns

Scheduled tasks

Next steps