AgentKit Studio
Guides

Guides

Fourteen task guides from creating an agent to shipping it, with copyable steps.

Guides are organized by what you want to accomplish. Each page gives prerequisites, copyable steps, a success check, and an error-code table. For the underlying model, read Core Concepts.

What this chapter contains

StagePageWhat you end up withPrerequisites
StartCreate and edit an agentA runnable agent with its first revisionAt least one configured model
ModelsConfigure models and credentialsSessions can select and actually call a modelAn OpenAI-compatible endpoint and API key
CapabilitiesBind an MCP serverThe agent can call tools from an external systemAn MCP endpoint or a launch command
CapabilitiesBind a skillThe agent reuses existing instructions and scriptsA directory containing SKILL.md
CapabilitiesBind knowledge bases and MemoryThe agent retrieves enterprise knowledge and remembers across sessionsA connected cloud control plane
RunRun and debug a sessionYou can read tool calls, file changes, and per-turn usageA working agent
GovernApprovals and permission policiesWrites and unknown side effects are constrainedA working agent
ShipDeploy to the cloudThe agent runs around the clockCloud credentials and an image registry
ShipManaged runtime update and rollbackUpdates are controlled and reversibleA deployed managed runtime
AutomateAutomation and schedulingRuns trigger on a schedule or an eventA deployed agent
ConnectChannelsExternal message entry points trigger the agentA deployed agent
ObserveObservability (OTel)Traces are exported over OpenTelemetryA deployed agent
RuntimesInstall, upgrade, and bind Star HarnessTake harness from "not enabled" to creatableNode.js and npm available

The three steps of the creation form

Creating and editing an agent share one form whose left-hand stepper is fixed at three steps (src/client/pages/agent-form.mjs:79):

StepWhat you decideRequiredCan you change it later?
Basic informationAgent ID, name, runtime, purpose, system prompt, task descriptionAgent ID and nameEverything except the runtime
Bind capabilitiesDefault model, available models, tools, MCP servers, skills, platform resourcesDefault modelYes, saving creates a new revision
Policy and confirmationAgent tool policy, cross-session memory, configuration summaryNoneYes

The runtime cannot be changed after creation — the runtime radio group is disabled in edit mode (src/client/pages/agent-form.mjs:266). To switch runtimes, create a new agent.

The five bindable resource kinds

The resource library and the creation form read the same catalog contract (src/resource-catalog.mjs:74):

kindSourceWhat the agent gets after bindingHow many per kind
modelLocal model profiles + model gatewayA switchable model list, one of them defaultMany, one default
mcpLocal MCP store + cloudEvery tool that server discoveredMany
skillInstalled skills + cloud Skill SpaceInstructions and bundled scriptsMany
knowledge-baseAgentEngine agentengine-serverRead-only retrievalOne
memory-instanceAgentEngine agentengine-serverCross-session recall and writesOne

Local skills and cloud Skill Spaces are managed separately, and both can be bound at the same time (src/client/pages/platform-resource-picker.mjs:12).

Defaults at a glance

These defaults come from code, not from UI copy. Before changing configuration, confirm which layer you are overriding:

ItemDefaultEvidence
Runtime for a new agentdsh-native (shown as DSH Native Harness)src/authoring-defaults.mjs:4
Selectable runtimesdsh-native / harness / codexsrc/client/theme.mjs:26
Runtime display namesDSH Native Harness / Star Harness / Codex Agentsrc/client/theme.mjs:28-33
Agent tool policystrictsrc/client/pages/agent-form.mjs:54
Execution policyworkspace-writesrc/execution-policy.mjs:10
Memory write modeexplicit_only on create, candidate on editsrc/client/state/agent-memory-config.mjs:76
Memory write switchshadow (evaluate only, never write)src/client/state/agent-memory-config.mjs:33
Default model credential nameAGENTKIT_MODEL_API_KEYsrc/client/dsh/model-bridge.mjs:17

Star Harness must be installed and bound first

Star Harness is the only pluggable one of the three runtimes: install it on the plugin page, enable it, and bind a workspace, and only then does the Agent form offer it (creatableRuntime, src/client/pages/agent-form.mjs:78). See Install, upgrade, and bind Star Harness.

How every page is structured

All guides follow the same skeleton so you can jump straight to the section you need:

  1. Prerequisites — what must already be in place; missing one makes a later step fail.
  2. Steps — numbered, with complete commands, paths, and field names. No ellipses.
  3. What success looks like — how to tell the step really worked, rather than "it did not error".
  4. Common errors — error code → cause → fix, searchable by code.
  5. Related reference — the matching contract page and source entry point.

Errors that cross pages

Error codeWhere it appearsCauseFix
AGENT_EXPECTED_REVISION_CONFLICTSaving an agentAnother window already saved a newer revisionClick Load latest version, verify, then save again
AGENT_NOT_FOUNDEditing, bindingThe agent was deleted or is outside this workspaceRefresh from the Agents list
AGENT_DEFINITION_REQUIREDSaving an agentThe submitted spec is emptyCheck whether you saved before the form finished loading
MODEL_CATALOG_UNAVAILABLEFirst step of capability bindingThe model gateway catalog failed to loadOnly configured profiles are shown; check credentials in Settings
MANAGED_RESOURCE_INVALIDBinding platform resourcesThe reference does not satisfy agentengine.resources/v1Re-select the resource instead of typing an ID
RESOURCE_CATALOG_UNAVAILABLEResource libraryOne catalog kind failed entirelyOther kinds still work; click Refresh to retry

All 188 error codes are listed in the error code reference.

Suggested reading order

Get one short path working first

Start with Quickstart and finish install → configure a model → create an agent → send a message. Come back to the guides only after all four succeed.

Then add capabilities one at a time

Add one resource kind, then immediately send a message that uses it. Adding MCP, skills, and a knowledge base at once makes failures impossible to attribute.

Each addition produces a new revision; Run and debug a session lets you compare turns.

Only then think about shipping

Working locally does not mean cloud resources are available. Deployment, managed runtimes, knowledge bases, and memory all depend on the AgentEngine control plane and credentials — see How the product relates to the cloud.

What you want to knowWhere to go
What a revision, build, and run freezeAgent · Revision · Build · Run
How the three runtimes differThe three runtimes
The shape of a resource bindingResources and bindings
All 134 control-protocol operationsControl protocol
Environment variables and configuration keysConfiguration reference
Troubleshoot by error codeTroubleshooting

On this page