Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
During Build or reuse, implement the declarative-agent component selected in your component plan. This article helps you choose the product procedure and confirm the component's readiness. You finish with a working agent, its dependencies, and test evidence for plugin integration.
You don't need an agent for every plugin. For your own orchestration, models, hosting, or runtime, follow the app path instead.
Check an existing agent before creating one
Review the existing agent's intended users, supported experiences, instructions, knowledge, actions, permissions, and known limitations with its owner.
Verify: Its documented capabilities match the task and target experience in your component plan.
Select the product's documented reuse or extension route. You can start with an agent template or copy an Agent Builder agent to Copilot Studio when that workflow fits.
Verify: The route supports your source and destination. Don't assume that adding an agent component to a plugin imports an existing project.
Agree on ownership, versioning, support, and change management, and then test representative tasks.
Verify: Record results and unresolved gaps before deciding to use the agent unchanged, configure it, or build a new component.
Start with Work IQ Dev Tools
For professional development, start with Work IQ Dev Tools, choose a documented environment, and add Agents Toolkit only when your component needs its specific project controls.
| Your starting point | Product procedure | Expected result |
|---|---|---|
| A new plugin that needs an agent component | Build a plugin with Work IQ Dev Tools | wiqd plugin create creates the container; wiqd plugin add agent scaffolds the optional agent component. |
| An existing declarative-agent source project | Agent component lifecycle | Continue agent-specific authoring and iteration with the documented wiqd agent commands. |
| Agent instructions, knowledge, or actions that need editing | Declarative agents in WIQD and the command reference | Saved source configured for the selected component and environment. |
Use the WIQD coding-assistant plugin for guided work or the direct CLI for repeatable commands. Install the coding-assistant plugin and CLI as separate prerequisites when needed. GitHub Copilot helps author files, instructions, configuration, and tests; it doesn't replace the authoring tool, runtime, or component requirements.
For an existing agent project, select offline validation explicitly:
wiqd agent validate --mode source --env dev
Verify: Read the saved-source findings and fix reported errors. A passing source check doesn't establish runtime behavior, endpoint reachability, authentication correctness, or final-package acceptance. The default package mode consumes an existing ZIP, uploads it for validation, and might require sign-in.
To add a skill to an existing WIQD agent project, check the agent-skills feature flag and current schema requirements in the agent skill command reference. A top-level plugin skill and a skill attached to a declarative agent are different registrations; neither establishes support in every host.
Use another product route when it fits the component
| Requirement | Product procedure |
|---|---|
| Managed authoring in Microsoft 365 Copilot | Agent Builder overview and Build agents |
| Low-code logic, managed environments, or Power Platform connectors | Copilot Studio and Connect and configure an agent for Teams and Microsoft Copilot |
| Lower-level agent manifests and supported development integrations | Work IQ Dev Tools (recommended for supported pro-code scenarios) |
| Skills in an existing Toolkit agent | Add skills with Agents Toolkit |
Standalone Copilot Studio agents appear in the Agent Registry, not under plugins. Don't assume that publishing a standalone agent repackages it as a plugin.
Configure and test the agent
For detailed implementation steps, follow the declarative-agent product guidance:
| Work to complete | Authoritative guidance | Verify |
|---|---|---|
| Define identity, behavior, task boundaries, and conversation starters | Build agents or Build with JSON, then Write effective instructions | Representative prompts produce the expected behavior and stay within the intended task. |
| Configure knowledge and capabilities | Knowledge sources and Add capabilities and custom actions | Required data is available to the intended user and unavailable when permissions don't allow it. |
| Add supported MCP or API actions | MCP and API plugins for declarative agents | Tool selection, authentication, consent, confirmations, errors, and unsupported requests behave as intended. |
| Add a supported skill | Skills in declarative agents and the selected authoring procedure | The target agent loads the skill and completes the defined task within that product's limits. |
| Manage configuration for your authoring route | Toolkit environments and versions or WIQD environments | Identity values, dependencies, and settings belong to the intended test environment. |
Record the agent version, owner, supported experiences, dependencies, known limitations, and test results. Product-specific testing is required; passing source or package validation isn't evidence that every user can run the task.
For design guidance, see Best practices for declarative agents and How the Copilot orchestrator chooses actions. When the required behavior works, you can optionally enable user feedback and optimize content retrieval.
Next step
Continue any remaining components in Build or reuse. When they're ready, enter Package & test with Integrate and test your components.