Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseInsight · On RadarGitHub Changelog→ original

GitHub Copilot Now Streams Agent Activity Logs to Microsoft Agent 365

Every trace of Copilot activity across a company can now be collected in one place. GitHub has started streaming Copilot's audit log into Microsoft Agent 365, Microsoft's control panel for AI agents. Administrators can see user sessions, including requests, model responses, and tool calls, along with usage trends. Risks are reviewed through Microsoft Purview and Defender, with no manual log assembly.

AI-processed from GitHub Changelog; edited by Hamidun News
GitHub Copilot Now Streams Agent Activity Logs to Microsoft Agent 365
Source: GitHub Changelog. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceGitHub Changelog ↗
Editor-in-Chief
Key metric #1Cloud agents, Copilot CLI, VS Code, Visual Studio, JetBrains, and Eclipse are all visible
Key metric #2The log includes prompts, responses, and tool calls for each session
Key metric #3Available on GitHub Enterprise Cloud with Enterprise Managed Users

When dozens or hundreds of developers use an AI assistant inside a company, the security team faces a simple question: what exactly are these assistants doing? What data do they receive, what do they answer, and which commands do they run on an employee's behalf? GitHub has answered with a new integration: Copilot audit activity can now be streamed directly into Microsoft Agent 365, a dashboard Microsoft positions as the control center for AI agents in an organization.

What was connected

Previously, Copilot events lived in the GitHub audit log. Now enterprise customers can send them to Agent 365 and view them alongside other AI workloads in the company. This matters because agents rarely stay within a single product: one team uses Copilot, another has a chatbot on its own model, and a third runs automation on third-party agents. A unified dashboard removes the need to piece together the picture by hand from several systems.

Coverage is broad. The stream will include activity from cloud agents on GitHub.com, Copilot CLI, Visual Studio Code, and Visual Studio. In addition, partner development environments are supported, such as JetBrains and Eclipse. In other words, it doesn't matter which editor a person prefers: their activity will still leave a trace in the log.

What the administrator sees

Agent 365 lets you review user sessions: prompts, model responses, and tool calls. This is the most valuable part. Knowing that an employee "used Copilot for 40 minutes" is one thing; seeing that an agent called external tools during a session and what requests it was sent is another. For example, if a developer pasted a configuration fragment containing sensitive data into a prompt, that can be spotted when the session is reviewed.

The dashboard also provides usage trends, so you can assess how actively teams use Copilot and in which clients.

Microsoft Purview and Defender are connected for investigations. Through them, administrators get additional risk information, can examine suspicious cases, and can take action. In effect, Copilot activity falls under the same security perimeter where companies already monitor email, documents, and endpoints, provided they use the Microsoft ecosystem.

Limitations and what to do in practice

The feature has requirements. It is available only to GitHub Enterprise Cloud customers with Enterprise Managed Users, meaning environments where employee accounts are managed by the company itself. If a team works with personal accounts or on other plans, it cannot get this stream. Setup is done through the enterprise-level audit log streaming mechanism, and GitHub refers to its documentation on the topic.

The published description does not say what the integration's limits are, whether there are delays in event delivery, or how long data is retained. These points should be clarified before enabling it. It also makes sense to plan privacy questions in advance: the log contains prompts and responses, so access to them should be restricted to a narrow group of people and agreed with legal counsel and company policy.

What teams should do:

  • If you are already on Enterprise Cloud with Managed Users, check the audit streaming settings and decide who will review sessions.
  • Define in advance which scenarios you consider risky: secret leaks in prompts, unwanted tool calls, unusual spikes in activity.
  • If you have a mixed set of AI tools, assess how far Agent 365 can serve as a single observation point.

The overall takeaway is straightforward: enterprise AI assistants are gradually getting the same level of control as other work systems. For developers, this means their interactions with agents become more transparent to security teams. For companies, it means scaling Copilot is easier when there is a clear log of actions and familiar investigation tools.

Story

Key Questions Answered

What was connected?

Previously, Copilot events lived in the GitHub audit log. Now enterprise customers can send them to Agent 365 and view them alongside other AI workloads in the company. This matters because agents rarely stay within a single product: one team uses Copilot, another has a chatbot on its own model, and a third runs automation on third-party agents. A unified dashboard removes the need to piece together the picture by hand from several systems. Coverage is broad.

What the administrator sees?

Agent 365 lets you review user sessions: prompts, model responses, and tool calls. This is the most valuable part. Knowing that an employee "used Copilot for 40 minutes" is one thing; seeing that an agent called external tools during a session and what requests it was sent is another. For example, if a developer pasted a configuration fragment containing sensitive data into a prompt, that can be spotted when the session is reviewed.

What are the key details on Limitations and what to do in practice?

The feature has requirements. It is available only to GitHub Enterprise Cloud customers with Enterprise Managed Users, meaning environments where employee accounts are managed by the company itself. If a team works with personal accounts or on other plans, it cannot get this stream. Setup is done through the enterprise-level audit log streaming mechanism, and GitHub refers to its documentation on the topic.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Hands-On AI Engineering · Hamidun Academy

Build & Ship Production AI Agents Without Boilerplate Code

Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.

Start 72h Free Trial →Vibe-Coding Intensive
What do you think?
Loading comments…