MC1476977 Low

Skills in Declarative Agents becoming generally available

  • User
  • Admin
  • Compliance
  • Copilot License

Summary AI-generated

Skills are now supported for Microsoft 365 Copilot Declarative Agents, allowing modular, reusable agent capabilities with secure execution and maintained governance.

Suggested actions AI-generated

  • Review Skills capabilities for Declarative Agents
  • Inform creators and users about the new capability
  • Update governance documentation if applicable
  • Monitor Microsoft documentation for guidance

Written by Azure OpenAI (gpt-4.1) from the text of the post below. It can be incomplete or wrong; the original post is authoritative.

Original post from Microsoft

[What and Why:]

Microsoft is introducing support for Skills in Declarative Agents. Skills are modular, reusable capabilities that package instructions, scripts, templates, and related resources into a reusable unit that can be added to agents. This helps organizations create more scalable, accurate, and maintainable agent experiences while supporting secure execution, governance, and enterprise-ready AI workflows.

[Rollout Schedule:]

  • Frontier: Available now.
  • General Availability (Worldwide): Rollout begins mid-September 2026.
  • Rollout completion: End of September 2026.

[Impact on Your Organization:]

Who is affected:

  • Organizations using Microsoft 365 Copilot Declarative Agents.
  • Admins deploying Declarative Agents.
  • Users creating their own Declarative Agents.
  • Organizations using Information Barriers are not supported at this time.

Platforms/Services:

  • Microsoft 365 Copilot
  • Declarative Agents
  • Web and desktop experiences supporting Declarative Agents

What will happen:

  • Skills can be added to user-created and admin-deployed Declarative Agents.
  • Skills package instructions, scripts, templates, and related assets into reusable modules.
  • Skills load only when needed, helping improve context efficiency and response quality.
  • Organizations can reuse supported skills across multiple agents.
  • Script execution uses a secure sandbox.
  • MIP sensitivity labels on skill files are retained and honored.
  • Existing governance controls continue to apply.
  • No changes occur unless creators choose to add skills.
  • Organizations using Information Barriers cannot use this capability currently.

[Action Required/Recommendations:]

No action is required.

  • Review Skills capabilities for your Declarative Agent strategy.
  • Inform Declarative Agent creators and users about the new capability.
  • Update internal guidance and governance documentation if applicable.
  • Monitor Microsoft documentation for additional implementation guidance.

[Compliance Considerations:]

  • Skills introduce new agent extensibility capabilities.
  • Users gain a new method of extending generative AI experiences.
  • Existing administrative governance continues to apply.

Additional details from Microsoft

Summary
Skills in Declarative Agents, modular reusable capabilities for Microsoft 365 Copilot, will be generally available worldwide by end of September 2026. They enhance scalability, security, and governance without requiring action. Organizations using Information Barriers are currently unsupported. Existing governance and compliance controls remain in effect.