Neo Code Agents =============== A **Neo Code Agent** is an agent you build and host yourself, registered into OneByZero Neo. Your own service runs the agent's logic. Neo passes each request to it and adds access control, versioning, and monitoring around it. .. Screenshot needed: the Remote Agent card on the Create Agent page (neo-code-agent-card.png) **Use a Neo Code Agent when you need:** - To bring in an agent that already exists — you have a working service and want to manage it alongside the agents built in Neo. - Logic that does not fit a visual flow — a specialised framework, your own libraries, or code your team prefers to keep in its own repository. - Control of your own release cycle — you decide the dependencies, the deployment, and when the agent changes. If you are starting from scratch, one of the in-platform builders will be quicker: the :doc:`Chatbot Builder ` for rule-guided conversations, the :doc:`Agentic Flow Builder ` for multi-step reasoning, or the :doc:`Voice Agent Builder ` for phone calls. How a Neo Code Agent Differs ---------------------------- .. list-table:: :widths: 28 36 36 :header-rows: 1 * - Aspect - Agentic Flow agent - Neo Code Agent * - Where the logic runs - Neo's flow engine - **Your own service** * - Visual canvas - Yes — you build the graph - **None** — there is no Workflow tab * - How you change behaviour - Edit the flow and republish - Change and redeploy your own code * - Testing in the platform - Playground - **No Playground** — test through NeoGPT or the run API * - Guardrails and approvals - Applied by the flow engine - Handled in your own service * - What Neo provides - Runtime, orchestration, tools - Registration, access control, versioning, monitoring Where Your Agent Runs ---------------------- Every Neo Code Agent uses one of two deployment types. You choose when you create it, and you can change it later. .. list-table:: :widths: 22 78 :header-rows: 1 * - Deployment type - What it means * - **External URL** - You host the service wherever you like and give Neo the address. You look after uptime, scaling, and networking. * - **EKS Deployment** - You supply a container image and Neo runs it on the platform's cluster. Neo works out the address for you once the deploy succeeds. See :doc:`creating` for the fields each option asks for, and :doc:`deployment` if Neo will be hosting your container. Prerequisites -------------- .. list-table:: :widths: 32 68 :header-rows: 1 * - What you need - Detail * - The right permissions - **Create Remote Agent** to add one, **Edit Remote Agent** to change it afterwards. See :doc:`permissions`. * - Your agent service - For **External URL**, the service must already be running and answering the endpoint described in :doc:`service_contract`. * - A container image - For **EKS Deployment**, push your image to AWS ECR and have the IAM role ARN ready. See :doc:`creating`. * - Network access - Neo calls your service directly, so check that firewall rules and security groups allow it. Neo sends no credentials with the request, so secure the endpoint yourself — see :doc:`service_contract`. Deepdive --------- .. list-table:: :widths: 32 68 :header-rows: 1 * - Page - What it covers * - :doc:`Creating a Neo Code Agent ` - Add the agent and fill in its details, for either deployment type. * - :doc:`The Service Contract ` - The one endpoint your service must answer, and what Neo sends it. * - :doc:`EKS Deployment ` - Deploy, redeploy, and undeploy a container Neo hosts for you. * - :doc:`Settings and Publishing ` - Output fields, KPIs, execution policy, and publishing a version. * - :doc:`Running and Monitoring ` - Run the agent, then follow its conversations, analytics, and traces. * - :doc:`Permissions ` - Who can create, edit, deploy, and monitor a Neo Code Agent. .. toctree:: :hidden: creating service_contract deployment settings_and_publishing monitoring permissions