Claude Code Gets Smarter: Self-Hosted Runners, ZIP Plugins & Opus 5 with 1M Context
Claude Code's latest update introduces enterprise-grade capabilities with Self-Hosted Runners, ZIP-based plugin installation, and Claude Opus 5's 1M-token context window. This article explores how these features improve security, extensibil
Claude Code Gets Smarter: Self-Hosted Runners, ZIP Plugins & Opus 5 with 1M Context
AI coding assistants are evolving quickly, and the latest Claude Code update introduces features that make development more flexible, secure, and scalable. Whether you're an individual developer or part of an enterprise engineering team, these improvements can significantly enhance your workflow.
In this article, we'll explore three of the biggest additions:
- Self-Hosted Runner
- Archive Plugin Source
- Claude Opus 5 with a 1 Million Token Context Window
1. Run Claude Code on Your Own Infrastructure
One of the most significant additions is the Self-Hosted Runner.
Instead of relying entirely on Anthropic's managed execution environment, developers can now run Claude Code on their own infrastructure using:
claude self-hosted-runner
This allows organizations to execute Claude Code inside their own:
- Docker containers
- Virtual machines
- Kubernetes clusters
- Private cloud environments
- On-premise servers
Why it matters
Many companies cannot expose internal repositories or production environments to external services due to compliance or security requirements.
The Self-Hosted Runner solves this by allowing Claude Code to execute entirely within your infrastructure while still leveraging Claude's reasoning capabilities.
Benefits
- Improved security
- Access to private repositories
- Internal API connectivity
- Compliance with enterprise policies
- Better control over execution environments
This makes Claude Code much more practical for enterprise software development.
2. Install Plugins Directly from ZIP Archives
Claude Code has also introduced a new Archive Plugin Source.
Previously, plugins were primarily installed from Git repositories or npm packages. Developers can now distribute plugins as ZIP archives hosted over HTTPS.
For example:
https://example.com/my-plugin.zip
Why this is useful
Not every organization wants to publish plugins through Git or npm.
Many internal developer tools are private.
With ZIP-based installation, companies can:
- Host plugins on internal servers
- Share plugins without Git access
- Package plugins for customers
- Maintain private enterprise extensions
Added Security
Anthropic also supports optional SHA-256 checksum verification.
This ensures the downloaded plugin has not been modified or tampered with before installation.
This simple addition makes plugin distribution significantly easier while improving software integrity.
3. Claude Opus 5 Becomes the Default Opus Model
The headline feature of this release is Claude Opus 5.
It replaces the previous default Opus model and introduces substantial improvements for software engineering tasks.
Key Improvements
- 1 million token context window
- Better reasoning
- Stronger agent workflows
- Improved long-running coding sessions
- Better multi-file understanding
- Enhanced tool usage
Unlike previous versions, the 1M context window is now the default, allowing developers to work with extremely large codebases without constantly trimming context.
What Does a 1 Million Token Context Mean?
A larger context window allows Claude to understand much more information during a single session.
Instead of loading only a few files, Claude can keep track of:
- Entire repositories
- Architecture documentation
- API specifications
- Design documents
- Multiple services
- Large conversations
This greatly reduces context switching and improves consistency during long development sessions.
For developers working on monorepos or enterprise applications, this is a major productivity improvement.
Better for AI Coding Agents
Claude Opus 5 is designed specifically for complex development workflows.
It performs better at:
- Large-scale refactoring
- Architecture planning
- Multi-step debugging
- Repository-wide code generation
- Long-running autonomous coding tasks
Teams building AI-powered development pipelines or autonomous coding agents will especially benefit from these improvements.
Enterprise Impact
These three features work together to make Claude Code much more enterprise-ready.
| Feature | Benefit |
|---|---|
| Self-Hosted Runner | Keep execution inside your infrastructure |
| Archive Plugin Source | Easier private plugin distribution |
| Opus 5 (1M Context) | Understands much larger codebases |
Together they improve:
- Security
- Flexibility
- Developer productivity
- Large-scale AI-assisted software development
Final Thoughts
Claude Code continues to evolve from a simple coding assistant into a full-fledged AI software engineering platform.
The introduction of Self-Hosted Runners gives organizations greater control over where code executes, Archive Plugins simplify distributing custom tooling, and Claude Opus 5's massive 1-million-token context enables developers to tackle large, complex projects with far less friction.
For teams building enterprise applications, internal developer platforms, or AI-assisted engineering workflows, these updates represent an important step toward making AI coding assistants practical for real-world production environments.
References
- Claude Code Changelog: Claude Code GitHub Changelog
- Claude Code Releases: Claude Code Releases
- Claude Opus 5 Documentation: Claude Opus 5 – What's New
- Claude Code Documentation: Claude Code Docs
Sources: The information in this article is based on Anthropic's official Claude Code changelog and documentation. (github.com)
