Pick the right mode for you in one minute — a getting-started overview for new users.
Qoder Cloud Agents is Agent as a Service. Enterprises no longer need to build a full agent infrastructure themselves — from building, deploying, and running an agent, to API integration, IM channel reach, and identity isolation, everything is delivered as a fully managed service by Qoder. It shortens the distance from "developing an agent" to "actually serving end users" from weeks to hours, so every team can quickly own its AI agent product.
The public API consists of two modes that cover the full path from building an agent to delivering it. Each mode exposes its own resource-management endpoints; there is no separate Resources layer.
A lower-barrier way to bring agents into business scenarios. Based on a three-tier configuration model (enterprise / template / identity), an administrator presets the agent shape and available resources, and callers use it out of the box. It also ships with the surrounding ecosystem that business needs — IM channel integration, scheduled tasks, and end-user identities.
The atomic mode that provides fully managed agent definition and runtime. No need to build your own agent loop or tool-execution sandbox — just define an agent via API, start a session, and dynamically mount the sandbox environment, skills, files, and other resources at session start to run complex tasks in the cloud and receive results in real time. Managed Mode also contains the organization-level APIs for Environments, Skills, Vaults, Files, Memory Stores, and Models.
If your goal is to quickly deliver agent capabilities to end users or business systems, start with Forward Mode. If you need full control over the agent's runtime behavior and to build your own upper-layer product logic, use Managed Mode. Use the resource endpoints under the selected mode; Forward and Managed resources are maintained independently.
About Qoder Cloud Agents
Qoder Cloud Agents is Agent as a Service. Enterprises no longer need to build a full agent infrastructure themselves — from building, deploying, and running an agent, to API integration, IM channel reach, and identity isolation, everything is delivered as a fully managed service by Qoder. It shortens the distance from "developing an agent" to "actually serving end users" from weeks to hours, so every team can quickly own its AI agent product.
The public API consists of two modes that cover the full path from building an agent to delivering it. Each mode exposes its own resource-management endpoints; there is no separate Resources layer.
Forward Mode
A lower-barrier way to bring agents into business scenarios. Based on a three-tier configuration model (enterprise / template / identity), an administrator presets the agent shape and available resources, and callers use it out of the box. It also ships with the surrounding ecosystem that business needs — IM channel integration, scheduled tasks, and end-user identities.
Managed Mode
The atomic mode that provides fully managed agent definition and runtime. No need to build your own agent loop or tool-execution sandbox — just define an agent via API, start a session, and dynamically mount the sandbox environment, skills, files, and other resources at session start to run complex tasks in the cloud and receive results in real time. Managed Mode also contains the organization-level APIs for Environments, Skills, Vaults, Files, Memory Stores, and Models.
Forward Mode vs Managed Mode
| Dimension | Forward Mode | Managed Mode |
|---|---|---|
| Positioning | Wraps a three-tier config model and IM / scheduling / identity capabilities on top of Managed Mode, helping teams deliver agents to end users quickly and reliably | Provides fully managed agent definition and runtime, exposing atomic runtime APIs |
| Best for | SaaS vendors, integrators, and teams serving C-end users or a large number of callers | Developers / enterprises with strong engineering capability who want full control of the agent shape and to build their own upper-layer product |
| Config model | Three-tier (enterprise / template / identity); admin presets, callers pass only template_id + identity_id | Callers explicitly specify environment / skill / files at session start |
| Caller complexity | Lower — Forward Mode absorbs the business complexity | Higher — very flexible; each call decides what to mount |
| End-user identity | Built-in Identity; one identity per C-end user, with automatic memory and permission isolation | Not built in; callers manage isolation themselves |
| IM channels | Built-in Lark / DingTalk / WeChat / WeCom, QR-code binding | Implement your own |
| Scheduled / triggered execution | Built-in Schedules (cron / one-time), effective once configured | Orchestrate via Managed Mode Deployments |