Banuba Launches Agent Skills for Video & Photo Editor SDKs

Banuba has officially released Agent Skills for its Video Editor SDK and Photo Editor SDK, marking a shift from traditional feature-building to outcome-based development. The announcement came via Business Wire on April 30, 2026, positioning the company at the intersection of augmented reality and AI-assisted development.

Agent skills function as portable knowledge packs that install directly into AI coding environments. Developers describe what they want to build in natural language, and the AI orchestrates the implementation using Banuba's underlying SDK capabilities. This eliminates the need to manually wire features or memorize API endpoints.

"Agent skills represent a shift from building features to describing outcomes," said Anton Liskevich, CPO and Co-founder at Banuba. "Developers can now go from idea to application faster than ever, without deep technical overhead."

The technical implementation is more nuanced than the marketing suggests. Agent skills bundle offline documentation, guided code generation, and autonomous scaffolding into a single package. When a developer invokes a skill via slash command in their coding assistant, the system loads context-specific information rather than dumping an entire SDK reference into every prompt. This keeps the AI's attention focused on the actual file being edited.

Three distinct skills ship with the launch: build-ve for Video Editor SDK projects, build-pe for Photo Editor SDK projects, and explain-ve-pe-docs for documentation lookups. The build skills include starter kit templates that detect the project platform and generate code accordingly. A developer can type "/build-ve Set up a Video Editor for Android with AI Clipping" and receive a complete scaffolded project.

Support extends across Claude Code, Codex, and Qwen Code. Installation happens through the Vercel Skills CLI with commands like "npx skills add @banuba/agent-skills -a claude-code". The skills load into platform-specific directories (.claude/skills/, .codex/skills/, .qwen/skills/) and activate automatically.

Platform coverage includes Android, iOS, Flutter, and React Native. Both SDKs are at version 1.51.0 as of the release. The agent skills are versioned and pinned to specific SDK releases, ensuring generated code matches the API being integrated against. This prevents the common problem where AI-generated code references deprecated methods or missing endpoints.

The company's blog post provides additional context on the integration workflow. Banuba's official documentation explains that neural networks guide users through every step, including setting up the development environment and obtaining necessary tokens. For experienced developers, the skills offer higher accuracy when answering SDK-related questions and platform-specific integration options.

Face AR SDK agent skills are in final testing stages and will release to the public soon. This suggests Banuba is rolling out the capability incrementally rather than all at once. The phased approach makes sense given the complexity of face tracking and virtual try-on technologies.

From a business perspective, this move lowers the barrier to entry for building advanced content creation experiences. Small teams or individual developers who previously couldn't justify the time investment in SDK integration can now prototype applications in minutes. The physical reality of this change matters: instead of spending hours reading documentation, clicking through API references, and debugging integration errors, developers type a prompt and wait for the scaffolding to appear.

There's a tradeoff here. The convenience of natural language prompts means less direct control over implementation details. Developers who need fine-grained customization may find themselves fighting the AI's assumptions about what "standard" means. The agent skills optimize for common use cases, not edge cases.

Banuba has been in the augmented reality space for over a decade, pioneering face tracking, virtual try-on, and virtual background technologies. This announcement extends that expertise into the AI coding assistant ecosystem. The company isn't just providing SDKs anymore; it's providing SDKs that AI assistants understand natively.

The GitHub repository for the agent skills shows the technical architecture in detail. The official repository contains portable skill definitions in the .agents/skills/ directory, with platform-specific packaging in subdirectories for each coding assistant. Each skill includes a SKILL.md file defining its behavior and capabilities.

Unlike passive documentation dumps, these skills carry executable behavior. They can apply starter-kit templates, wire up dependencies, and scaffold entire projects end-to-end. This distinction matters when you're actually building something. A documentation dump answers questions; an agent skill builds the thing you asked for.

Installation is straightforward but requires access to the Vercel Skills CLI. Developers need to run the appropriate command for their coding environment, then invoke skills with slash commands. The workflow is: install skills, invoke with slash command, review generated code, customize as needed. It's not magic, but it's significantly faster than starting from scratch.

The timing of this release aligns with broader industry trends toward AI-assisted development. Multiple coding assistants now support plugin architectures, and SDK providers are racing to optimize their tools for AI consumption. Banuba's approach of bundling offline documentation with executable scaffolding represents a more sophisticated implementation than simple API documentation injection.

For enterprise teams, the agent skills could accelerate prototyping cycles. A product manager with a rough idea can describe the desired outcome, have the AI scaffold a working prototype, and then hand off to engineers for refinement. This compresses the feedback loop between concept and tangible output.

Whether users actually pay for it remains the real question. The SDKs themselves are commercial products, and the agent skills are positioned as a value-add to existing integrations. The question is whether the time savings justify the cost for teams that already have established development workflows.

Some developers will find the agent skills genuinely useful. Others will see them as another layer of abstraction between them and the code. The technology works, but adoption depends on whether the workflow matches how teams actually build software. Time will tell if this becomes standard practice or remains a niche tool for rapid prototyping.

Banuba's next moves will likely include expanding agent skills to additional SDKs and refining the AI's understanding of edge cases. The Face AR SDK release is the immediate next step. Beyond that, the company may explore deeper integration with development pipelines or offer enterprise-specific customization options.

The agent skills represent a legitimate technical advancement in SDK distribution. They solve real problems around documentation freshness, context management, and scaffolding efficiency. Whether they become essential tools or interesting experiments depends on developer adoption and the quality of the generated code in production environments.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn