An abstract coding agent session timeline flows into a secure audit collector
An abstract coding agent session timeline flows into a secure audit collector
+ Microsoft News

GitHub exposes Copilot agent sessions for enterprise audit

GitHub's public preview lets enterprise owners stream or retrieve Copilot agent session data, including prompts, responses, and tool calls, across major Copilot clients.

GitHub opened Copilot agent session streaming in public preview on July 2 for GitHub Enterprise Cloud customers with enterprise managed users.

The preview gives enterprises access to Copilot agent session data across GitHub cloud agents on github.com and data-resident ghe.com, Copilot CLI, Visual Studio Code, Visual Studio, and partner IDEs including JetBrains and Eclipse. GitHub says the data includes session activity such as prompts, responses, and tool calls.

This is the governance half of agentic coding. Once an assistant can inspect code, run tools, call models, and create changes across multiple clients, enterprises need more than a usage chart. They need a record of what the agent did.

Session data closes a visible audit gap

GitHub’s own Copilot audit-log documentation says the enterprise audit log records changes to Copilot plans, settings, licenses, and agent activity on the GitHub website. It also says the audit log does not include client session data such as prompts sent locally to Copilot.

The new preview changes that boundary for eligible enterprises. GitHub says companies can stream session data to an event collector or SIEM from audit-log settings. Microsoft Purview is also available as a supported streaming endpoint in public preview.

There is a REST path as well. Enterprise owners can retrieve the last 48 hours of session data on demand through GET /enterprises/{enterprise}/copilot/usage-records.

That 48-hour pull window matters. The streaming path looks like the durable compliance route. The REST API is useful for checks, incident response, and backfill, but it is not a substitute for long-term retention if an organization needs audit history.

Agent governance is becoming operational

This release lands one day after GitHub posted a cluster of Copilot changes around model routing, AI-credit limits, browser tools, vision, and enterprise-managed defaults.

The pattern is coherent. GitHub is adding more agent capability and more enterprise control at the same time. Model routing lets organizations govern what models are used. Credit limits put a session boundary around spend. Session streaming gives security and platform teams a record of activity.

That does not remove the need for policy. It makes policy enforceable and inspectable. If a developer uses a browser tool, a cloud agent, or an IDE agent to touch sensitive code, the organization now has a path to collect evidence across the major Copilot surfaces.

The preview is narrow by design

The availability line is important. This is not every Copilot user. GitHub names GitHub Enterprise Cloud customers with enterprise managed users.

That scope fits the feature. Session records can include prompts, responses, tool calls, and repository context. The buyers most likely to turn this on first are regulated companies, security-conscious engineering organizations, and large platform teams already sending audit logs into a SIEM.

Sources

The AI Feed Desk

The AI Feed Desk

Editorial desk

The AI Feed Desk tracks AI provider updates, model releases, agent tooling, and enterprise adoption, turning fast-moving announcements into source-linked context for builders and operators.

Noticed a typo, incorrect information, or translation error?

Tell us so we can fix it.

Help Improve This Article

Related Articles

A routing switchboard sends coding tasks into different model lanes from an admin control plate

Copilot auto model selection turns routing into an admin setting

GitHub Enterprise admins can now default Copilot conversations to auto model selection through managed-settings.json, tying model routing to enterprise AI standards.

The AI Feed Desk

By The AI Feed Desk

Shared AI credit tokens flow through team lanes while one cost center lane is stopped by a cap gauge

GitHub cost centers can cap shared AI credit pools

GitHub added AI credit pools for cost centers, letting enterprise admins cap how much monthly included Copilot credit one group can draw before metered budgets apply.

The AI Feed Desk

By The AI Feed Desk

Developer tasks pass through a policy gate into several abstract model routing lanes with a cost meter below

GitHub turns Copilot model choice into a governed routing layer

GitHub's July 1 Copilot changes put model routing, browser tools, vision, Kimi K2.7 Code access, and AI-credit caps under clearer admin control.

The AI Feed Desk

By The AI Feed Desk

A usage dashboard shows AI credit consumption bars for individual developers

GitHub adds per-user AI credit metrics for Copilot admins

GitHub's Copilot usage metrics API now reports per-user AI credit consumption, giving enterprise and organization admins a clearer signal for adoption and budget planning.

The AI Feed Desk

By The AI Feed Desk

A browser-shaped glass panel, image tile, and testing checklist converge on a developer workstation

Copilot browser and vision tools make agents multimodal testers

GitHub made Copilot browser tools in VS Code generally available and opened Copilot vision across plans, giving coding agents browser, screenshot, image, and PDF context.

The AI Feed Desk

By The AI Feed Desk