GitHub Release Notes

GitHub Copilot with OpenCode

GitHub is officially supporting using your Copilot Pro, Pro+, Business, or Enterprise subscription with OpenCode.
GitHub Copilot now fully supports authentication with OpenCode through a formal partnership, allowing you to use your existing Copilot subscription across more of your development workflow.

What is OpenCode?

OpenCode is an open source agent that helps you write code in your terminal, IDE, or desktop. You can learn more at OpenCode’s site.

How it works

Getting started with OpenCode and GitHub Copilot is simple:

In OpenCode, run /connect and select GitHub Copilot.
Complete the GitHub device login flow.
Start using OpenCode powered by Copilot.

Who can use this feature

All developers with paid GitHub Copilot subscriptions (Pro, Pro+, Business, or Enterprise) can now authenticate into OpenCode using their Copilot credentials—no additional AI license needed.

Learn more about OpenCode’s Copilot provider setup.
Join the discussion within GitHub Community.

The post GitHub Copilot now supports OpenCode appeared first on The GitHub Blog.

Original source Report a problem

GitHub Actions cache now has a rate limit of 200 uploads/minute per repo. This limit only impacts uploads of new cache entries—it does not impact cache entries that are downloaded in job runs.

We’ve implemented this limit in response to repositories that upload a high volume of cache entries in very short periods of time which increase cache thrash and impact system stability. We’re working with the authors of some marketplace actions to use behavior that reduces these challenges.

Users impacted by the rate limits can expect some new cache entry uploads to be rejected until the rate limit resets.

The post Rate limiting for actions cache entries appeared first on The GitHub Blog.

Original source Report a problem

Copilot memory is now available in public preview for all paid GitHub Copilot plans

Copilot memory enables Copilot to learn and retain useful details about your repositories. As you work with Copilot coding agent, Copilot code review, or Copilot CLI, it builds a repository-specific understanding that improves the quality of its assistance over time.

How it works

Copilot automatically captures tightly scoped insights called “memories” as it works in your repositories. These memories are:

• Repository-specific and validated against the current codebase before use.
• Shared across Copilot features, so what coding agent learns can help code review.
• Automatically expired after 28 days to prevent stale information.

For more details on how exactly Copilot memory works, check out our engineering blog post on Copilot memory.

Get started

Individual users (Copilot Pro or Pro+): Enable Copilot memory in your personal Copilot settings.
Organizations and enterprises: Enable through policy settings in your organization or enterprise settings.
Repository owners can review and delete stored memories in Repository Settings > Copilot > Memory.
To learn more about how to use this feature, check out our Copilot memory documentation.
Join the discussion within GitHub Community.

The post Agentic memory for GitHub Copilot is in public preview appeared first on The GitHub Blog.

Original source Report a problem

What’s new

New provider options

You can now connect API keys from AWS Bedrock, Google AI Studio, and any OpenAI‑compatible provider. These options join Anthropic, Microsoft Foundry, OpenAI, and xAI as supported BYOK choices.

Support for the Responses API

BYOK now supports models that use the Responses API. This enables structured outputs and richer multimodal interactions.

Ability to set a maximum context window

Admins can define the maximum context window for BYOK models, helping balance cost, performance, and response quality.

Streaming responses for faster interaction

Streaming output from connected models lets Copilot display responses as they are generated, rather than waiting for completion.

Start using the new BYOK features today

These new capabilities are available now in public preview for GitHub Enterprise and Business customers. Connect your LLM provider’s API key in your enterprise or organization settings and start using your models in Copilot Chat and supported IDEs. Learn more by visiting our BYOK documentation.

Help us shape the future

We are just getting started and your feedback will guide what comes next. Join the discussion within the GitHub Community to share feedback and connect with other developers.

The post GitHub Copilot bring your own key (BYOK) enhancements appeared first on The GitHub Blog.

Original source Report a problem

Hierarchy view for GitHub Projects is now available in public preview

You can now view your full issue hierarchy directly in project table views, giving you clear visibility into complex work breakdowns without losing context or switching views.

🔗 What is hierarchy view?

Hierarchy view lets you see the full nesting of sub-issues directly in GitHub Projects.

With hierarchy view enabled, you can:

• Expand and collapse sub-issues up to eight levels deep.
• Group, slice, sort, and filter while preserving hierarchy.
• View and quickly add existing sub-issues that aren’t in the project yet.

Getting started

To enable hierarchy view, open the View menu and turn on Show hierarchy.

Coming soon

We’re continuing to invest in hierarchy view. Next up:

• The ability to create and add sub-issues inline, directly from the project.
• The ability to drag-and-drop reordering and reparenting of sub-issues.
• A synced order between the sub-issue list on issues and the sub-issue list in projects.

Tell us what you think

We’d love your feedback. Join the conversation in GitHub Community, or select Give feedback from the projects … menu to share your thoughts.

🚀 Additional improvements to issues and projects

Performance enhancements to issues

We’re continuing to make issues faster. This week, we shipped our first performance improvement focused on making recently viewed issues feel instant. Using load times of under 200ms as our benchmark for “instant,” this update increased the percentage of instantly loading issues from 2% to 12%.

This is the first of several performance improvements planned over the coming months to make working with issues consistently fast.

Other improvements

• We’ve added a new View menu in projects to make it easier to find view display options.
• There is a new REST API endpoint to create project views.
• Pull requests in the merge queue now display their merge queue state icon in projects.
• You can now paste a pull request URL from the “Development” section of an issue to quickly link items.

The post Hierarchy view now available in GitHub Projects appeared first on The GitHub Blog.

Original source Report a problem

GitHub secret scanning: extended metadata

GitHub secret scanning is adding support for extended metadata checks in security configurations, starting February 18, 2026.
Extended metadata checks are a nested feature under validity checks. If you’ve already enabled validity checks for your repositories via security configurations, you’ll automatically have extended metadata checks enabled.

What are extended metadata checks?

With extended metadata checks, secret scanning alerts now display details about a secret’s owner, secret creation and expiry dates, and project or organization context when information is available from the secret provider. This feature extends validity checks to additional context about the secret.
For example, leaked OpenAI keys with information available will display the secret owner’s name, email, and identifier, in addition to information about the organization.
These new metadata keys expand on existing validity checks to give more actionable context for triage and remediation, enabling development and security teams to assess exposure faster and prioritize remediation.
Note: The availability of metadata depends on the secret provider, the type of token, and sometimes even the secret itself. GitHub makes a best effort to display all available metadata, but not every key will always be present.

What’s changing?

Extended metadata checks are currently available to Enterprise Cloud customers with secret scanning who have validity checks enabled. Starting on this date, users will be able to enable or disable the feature at organization and enterprise levels with security configurations. Repositories with validity checks enabled with security configurations will see metadata checks automatically enabled for them at this time.

Learn more and share feedback

Learn more about securing your repositories with secret scanning or share feedback about secret scanning and extended metadata checks.

The post Secret scanning extended metadata to be automatically enabled for certain repositories appeared first on The GitHub Blog.

Original source Report a problem

New models and easier model management

As we wrapped up 2025 and rang in 2026, we’ve continued to deliver new ways of working with agents in GitHub Copilot CLI while improving the terminal-native experience for all developers. Here’s what’s new since our last update.

GPT-5 mini and GPT-4.1, models that are included with your Copilot subscription and do not consume premium requests on paid plans, are now available. Run /model to see all model options.
If you select a model that’s disabled in your policy settings, the CLI now prompts you to enable it directly—no need to leave the terminal. This works in the model picker, /model command, and --model flag for Copilot Pro/Pro+ users.

Built-in custom agents

Copilot CLI now includes specialized custom agents for common tasks:

• Explore: Fast codebase analysis. Ask questions about your code without cluttering your main context.
• Task: Runs commands like tests and builds. Receive brief summaries on success, full output on failure.
• Plan: Creates implementation plans by analyzing dependencies and structure.
• Code-review: Reviews changes with high signal-to-noise ratio, focused on only surfacing genuine issues.

Copilot delegates to these agents automatically when appropriate and can run multiple agents in parallel.
When combined with Agent Skills, you can more easily integrate agentic workflows into yoiur Copilot CLI experience.

New ways to install

Windows (WinGet):

winget install GitHub.Copilot

macOS and Linux (Homebrew):

brew install copilot-cli

macOS and Linux (install script):

curl -fsSL https://gh.io/copilot-install | bash

Package manager and install script installations automatically update.
Codespaces and dev containers: Copilot CLI is included in the default GitHub Codespaces image and available as a Dev Container Feature.
Standalone executables are also available in GitHub release artifacts for all platforms.

Automation and scripting

New flags make Copilot CLI easier to use in scripts and pipelines via copilot -p:

• Flag
  • --silent
    Suppress stats and logs for clean, parseable output
  • --share [PATH]
    Export session transcript to markdown file
  • --share-gist
    Export session to a shareable GitHub gist
  • --available-tools
    Allowlist specific tools
  • --excluded-tools
    Denylist specific tools
  • --additional-mcp-config
    Add MCP config files per-session

Refer to copilot --help for a list of all optional flags.

CI/CD authentication: Set GITHUB_ASKPASS to point to an executable that returns your token—useful for credential managers and pipelines.

Copilot Spaces: The GitHub MCP server now includes Copilot Spaces tools for project-specific context.

Context management

Auto-compaction: When approaching 95% of the token limit, Copilot automatically compresses your history.
/compact: Manually compress context anytime.
/context: Visualize token usage with a detailed breakdown.
--resume: Press TAB to cycle through local sessions and remote Copilot coding agent sessions.

Terminal experience

Better diffs: Intra-line syntax highlighting shows exactly what changed. It now integrates with Git’s configured pager.
Tab completion: Autocomplete paths in /cwd and /add-dir.
Ctrl+T: Toggle model reasoning visibility in supported models.
Cleaner history: Agent-run commands excluded from Bash/PowerShell history.

Web access controls

The new web_fetch tool retrieves content from URLs as markdown. URL access is controlled through ~/.copilot/config with allowed_urls and denied_urls patterns. These rules also apply to shell commands like curl and wget.

Share your feedback

Update with your package manager, or run npm install -g @github/copilot@latest.
Join the discussion in Copilot CLI’s public repository.
The post GitHub Copilot CLI: Enhanced agents, context management, and new ways to install appeared first on The GitHub Blog.

Original source Report a problem

Available SDKs

The Copilot SDK is available in four languages:

• Node.js / TypeScript: @github/copilot-cli-sdk
• Python: copilot
• Go: github.com/github/copilot-cli-sdk-go
• .NET: GitHub.Copilot.SDK

Key features

All SDKs provide a consistent API with:

• Multi-turn conversations: Maintain session history for context-aware interactions.
• Tool execution: Define custom tools that the model can invoke during conversations.
• Full lifecycle control: Manage client and session lifecycles programmatically.

Learn More

• SDK Documentation
• Node.js SDK
• Python SDK
• Go SDK
• .NET SDK

Share your feedback

• Join the GitHub Community to share your feedback.

The post Copilot SDK in technical preview appeared first on The GitHub Blog.

Original source Report a problem

Availability in GitHub Copilot

GPT-5.2-Codex is generally available to Copilot Enterprise, Copilot Business, Copilot Pro, and Copilot Pro+.

You’ll be able to select the model in the Copilot model picker from any of the following:

• Visual Studio Code in all modes: chat, ask, edit, agent
• Copilot Chat in github.com
• GitHub Mobile (iOS and Android)
• Copilot coding agent (Copilot Pro and Pro+ subscribers only)
• Copilot CLI

Rollout will be gradual. If you don’t see the new models yet, check back soon.

Enabling access

• Enterprise and Business plans: Administrators must enable the OpenAI GPT-5.2-Codex in Copilot settings. Once enabled, users in that organization will see the respective model in the Copilot model picker.
• Pro and Pro+ plans: Users can enable the model by selecting it in the model picker and confirming the one-time prompt.
• Bring your own key: Within Visual Studio Code, select Manage Models from the picker, choose OpenAI GPT-5.2-Codex, and enter your API key when prompted.

Learn more

To explore all models available in GitHub Copilot, see our documentation on models and get started with Copilot.

Share your feedback

Join the GitHub Community to share your feedback.

The post GPT-5.2-Codex is now generally available in GitHub Copilot appeared first on The GitHub Blog.

Original source Report a problem

Why organization custom properties

Organization custom properties are now generally available, giving enterprise administrators a powerful way to tag organizations with metadata and automatically target enterprise rulesets.

Enterprises often organize their GitHub presence around business units, geographic regions, compliance frameworks, or development stages. Until now, applying consistent rulesets meant manually selecting organizations one-by-one or relying on naming conventions that don’t scale.

With organization custom properties, you can:

• Define once, apply everywhere: Tag organizations with properties like region: EMEA, compliance: SOC2, or business-unit: Platform and let your rulesets do the rest.
• Eliminate configuration drift: Rulesets automatically apply to any organization matching your criteria, even newly created ones.
• Adapt to your enterprise structure: Whether you organize by department, regulatory requirements, or project lifecycle, custom properties meet you where you are.

Imagine you need to enforce stricter code review requirements for all organizations handling customer data in the EU. Instead of maintaining a manual list, create a ruleset that targets region: EU and data-classification: customer-data. When a new EU-based team spins up their organization, the policies apply automatically.

How it works

Enterprise administrators define custom properties at the enterprise level and assign values to individual organizations. Properties support:

• Single-select: Choose one value from a predefined list
• Multi-select: Choose multiple values from a predefined list
• Text: Enter free-form text
• True/false: Set a boolean value
• URL: Enter URL value

Once you’ve tagged your organizations, enterprise rulesets can target them based on any combination of properties.

Get started

Organization custom properties are available to all GitHub Enterprise Cloud customers.

Enterprise administrators can access custom properties in the organization section of their enterprise settings.

For more information, see “Managing custom properties for organizations in your enterprise” in GitHub Docs.

The post Organization custom properties now generally available appeared first on The GitHub Blog.

Original source Report a problem