Answer
Including a capability or dependency means explicitly declaring it in your toolchain—whether for AI agent actions, runtime environments, or build-time libraries. Builders now face more frequent inclusion decisions as native OS integration and real-time multimodal support become available out-of-the-box.
Key points
- Inclusion is a deliberate trade-off: added capability vs. increased surface area, latency, or maintenance.
- Native GUI interaction (e.g., Claude Code on macOS) and real-time multimodal processing (e.g., Qwen3.5-Omni) shift what must be included to ship certain workflows.
- Lightweight, DOM-free utilities like Pretext show that inclusion choices can also reduce overhead—especially in rendering- or measurement-sensitive contexts.
What changed recently
- Claude Code now includes native macOS GUI interaction via 'Computer Use' (March 31, 2026).
- Qwen3.5-Omni demonstrates production-ready real-time audio-visual programming (March 31, 2026); Pretext—a DOM-free TypeScript text measurement library—launched March 29, 2026.
Explanation
What you include directly affects observability, debugging scope, and deployment constraints. For example, enabling GUI interaction means your monitoring must capture OS-level events—not just API calls.
Recent releases lower the barrier to including previously complex capabilities—but don’t eliminate the need to audit side effects. Pretext’s 500× speed gain comes from excluding the DOM; Qwen3.5-Omni’s multimodal support requires including new input/output handlers.
Tools / Examples
- Including 'Computer Use' in a Claude Code agent means adding macOS accessibility permissions and logging GUI event traces.
- Including Pretext instead of DOM-based measurement avoids layout thrashing in screenshot-rendering pipelines.
Evidence timeline
Claude Code officially integrates 'Computer Use' capability, enabling native macOS GUI interaction; Qwen3.5-Omni fully demonstrates real-time multimodal capabilities across use cases including audio-visual programming, v
Pretext—a pure TypeScript text measurement library requiring no DOM—has been open-sourced, delivering a 500× performance boost and validated in real-world use cases including web screenshot rendering, generative UI (e.g.
Sources
FAQ
Do I need to include new capabilities just because they’re available?
No. Inclusion should follow a specific workflow need—not availability. Audit whether the capability reduces technical debt, enables a user outcome, or simplifies an existing workaround.
How do I verify what’s actually included in a model or library release?
Check release notes for declared interfaces and dependencies—not just marketing claims. RadarAI’s Signals Library documents observable behaviors, and methodology pages describe how inclusion claims are validated.
Last updated: 2026-03-31 · Policy: Editorial standards · Methodology