KAGENTXINTEL TERMINAL

Open Design

Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.

Who it is for: Solo Builder、Designer、Engineering、Product Managers、Marketing

Positioning

Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.

  • Product stage: The specified document records multiple versions and pricing pages up to August 18, 2026; the initial public release date is still to be verified.

Core features

  • Local files and project workspace: Desktop can read local files, Figma exports, and code repositories; the official FAQ states that desktop app, daemon, and skill runtime run on the user's machine, with generated outputs in the project directory.
  • Visual direction generation and sandbox preview: After setting the visual direction, the system generates real runnable files and previews and edits them in a sandbox.
  • Multiple outputs and export formats: The official website lists prototypes, landing pages, slide decks, dashboards, brand and design systems, and HTML video; README lists Prototype, Deck, Mobile app, Image, Document, HyperFrame, and HTML, PDF, PPTX, MP4 export formats.
  • Code agent compatibility: The official FAQ states that the product provides 17 first-party BYOK adapters and supports switching code agents.
  • Design systems, preferences, and memory: The official website states that design choices are consolidated into design systems, preferences, and memory, influencing subsequent generations.

Target users

  • Solo Builder: Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.
  • Designer: Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.
  • Engineering: Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.
  • Product Managers: Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.
  • Marketing: Open Design's official website describes it as an open-source, local design workspace: using the coding agents run by users to generate workflows from design intent to prototypes, websites, slides, and HTML videos.

Pricing

  • The specified document indicates Free, BYOK, local code agents, as well as Plus, Pro, Max model quotas, and Team per-seat billing paths. · Free is $0 / mo; Plus first month $16, then $20 / mo, annual $168; Pro first month $70, then $100 / mo, annual $720; Max first month $120, then $200 / mo, annual $1,176; Team shows $63 and $105 / seat / month, at least 3 seats. Open Design Cloud's hosted model charges based on actual usage; BYOK calls do not consume Open Design balance.: Revenue, number of paying users, churn rate, retention rate, quota consumption rate, financing, and enterprise client scale are unknown. Enterprise-specific pricing, contract boundaries, support content, and compliance commitments are unknown.

Verdict

The specified document indicates that Open Design's product mechanisms revolve around local operation, code agent access, design intent guidance, real file generation, and preview, and records multiple versions up to August 18, 2026. Reddit, X, and independent reviews require attributable user samples; although Zhihu, WeChat official accounts, and Xiaohongshu are marked as unavailable in evidence.feedback, the current ledger does not have separate claims for these three statuses. Revenue, retention, financing, enterprise client scale, and actual user satisfaction remain unverified, so no definitive investment or commercialization conclusion can be formed.

  • Strength: Local files, code repositories, code agents, and project directory delivery are integrated into the same product workflow. Product evidence covers design intent input, real runnable files, sandbox preview, and multiple export formats. The official website states that multiple agent adaptation and design systems, preferences, and memory mechanisms are provided. The official blog records multiple versions; the GitHub Releases page shows 0.19.2 as Latest, but the year of the page release is not displayed.
  • Limitation: Local daemon and code agent are current operational conditions. Public Issues show recovery, share link revocation, error diagnosis, and plugin preview problems, but the scale of issues and repair speed are not verified. Lack of attributable independent user feedback, revenue, retention, financing, and enterprise client scale evidence.

How it delivers value

  • Intent input: Users express design intent through natural language, reference images, or URLs, which are then generated into interfaces by AI. → Users express design intent through natural language, reference images, or URLs, which are then generated into interfaces by AI.
  • Context reading and generation: Agents read local files, Figma exports, and code repositories, and generate real runnable files based on the visual direction. → Agents read local files, Figma exports, and code repositories, and generate real runnable files based on the visual direction.
  • Preview and file delivery: The system previews and edits the generated outputs in a sandbox; README lists export formats such as HTML, PDF, PPTX, and MP4. → The system previews and edits the generated outputs in a sandbox; README lists export formats such as HTML, PDF, PPTX, and MP4.
  • Design system reuse: The official website states that design choices are consolidated into design systems, preferences, and memory, influencing subsequent generations. → The official website states that design choices are consolidated into design systems, preferences, and memory, influencing subsequent generations.

Competition

Current evidence can be used to compare product dimensions such as local operation, code agent compatibility, project file delivery, and runnable output generation; independent data for other products is not provided in the specified evidence.

  • vs Claude Design: Wins: The specified evidence supports Open Design's local operation, BYOK adaptation, and generated outputs in the project directory. Loses: The specified evidence does not provide functional, pricing, or user experience data for Claude Design, so specific outcomes cannot be confirmed.
  • vs Google Stitch: Wins: The specified evidence supports Open Design's ability to express design intent through natural language and generate real runnable files. Loses: The specified evidence does not provide functional, pricing, or user experience data for Stitch, so specific outcomes cannot be confirmed.
  • vs Figma: Wins: The specified evidence supports Open Design's ability to read Figma exports, local files, and code repositories, and deliver generated files. Loses: The specified evidence does not provide functional, pricing, or user experience data for Figma, so specific outcomes cannot be confirmed.
  • vs v0, Lovable, Bolt.new, etc., hosted generators: Wins: The specified evidence supports Open Design's local operation, BYOK, and generated outputs in the project directory. Loses: The specified evidence does not provide direct data for these products, so current features, pricing, and delivery quality cannot be compared.

Risks

  • Current operation requires at least a local daemon and one code agent, which may increase the initial usage threshold.: Current operation requires at least a local daemon and one code agent, which may increase the initial usage threshold.
  • Changes in multiple agents, model providers, and plugin interfaces may affect workflow stability, with the scale of impact unknown.: Changes in multiple agents, model providers, and plugin interfaces may affect workflow stability, with the scale of impact unknown.
  • GitHub Issues has listed daemon recovery, share link revocation, error classification, and Docker plugin preview issues, but the scale of issues and repair speed are not verified.: GitHub Issues has listed daemon recovery, share link revocation, error classification, and Docker plugin preview issues, but the scale of issues and repair speed are not verified.
  • BYOK calls, model quotas, and actual usage billing mechanisms may bring cost and predictability issues, with the degree of impact to be verified.: BYOK calls, model quotas, and actual usage billing mechanisms may bring cost and predictability issues, with the degree of impact to be verified.
  • Stars, contributors, plugins, and design systems numbers cannot replace active users, revenue, and retention metrics.: Stars, contributors, plugins, and design systems numbers cannot replace active users, revenue, and retention metrics.

Company

  • Stage: The specified document records multiple versions and pricing pages up to August 18, 2026; the initial public release date is still to be verified.
  • Funding: Not disclosed

User voices

  • Negative: The GitHub Issues page lists a problem where the macOS 0.19.2 desktop application exits midway in the daemon sidecar and cannot recover.
  • Negative: The GitHub Issues page lists a problem where public sharing links cannot be revoked after the daemon restarts.
  • Negative: The GitHub Issues page lists a problem where the Claude Code prompt-length error is incorrectly classified as an authentication failure, and the Docker image lacks the plugin-previews manifest.
  • Requested: The GitHub Issues page only provides publicly available individual issue reports and cannot represent overall user satisfaction, failure rate, or retention.
  • Requested: Reddit, X, and independent reviews require targeted collection for attributable user samples.
  • Requested: Zhihu, WeChat official accounts, and Xiaohongshu are marked as unavailable in evidence.feedback, but the current ledger does not have separate claims for these three statuses; therefore, they cannot be supported by C-002 for conclusions about these channels.
  • Requested: The number of official website indicators, GitHub stars, contributors, plugins, and design systems cannot directly prove user satisfaction, activity, or business success.

Sources

Updated 2026-08-18

Visit Open Design