v0.20.2

Open Science v0.20.2

A release about readability and reach: a complete Spanish interface, in-app sandboxed previews for source links in agent responses, a live variable browser for running notebook kernels, session hover previews and right-click preview-tab actions in the workspace, and new OpenCode Go and Zen providers plus GLM-5.3 family models — alongside faster conversation rendering, bounded save cadence for long sessions, and steadier archiving, plan preview, and persistence.

Open Science is an open-source, local-first AI research workbench for scientists and researchers. It enables reproducible, inspectable research across models with scientific AI agents, Python and R execution, scientific data connectors, and cross-platform support for macOS, Windows, and Linux.

v0.20.2 is a polish release. The interface is now fully available in Spanish — onboarding, settings, conversation surfaces, native dialogs, and release notes — joining the existing seven languages with a runtime switcher in Settings. Links in agent responses become native source links: hovering reveals the source title and full URL, and clicking loads the page in a sandboxed side-panel preview with a deterministic progress indicator, so reading a cited source no longer means leaving the workspace. Notebook gains a Variables view that inspects the running Python or R namespace — names, types, shapes, and bounded previews — read-only, refreshed after each execution, without starting a kernel just to browse. The sidebar shows a session's title and description on hover (with scrolling for overflowing titles), preview tabs get a right-click menu with Close, Close others, Download, Copy path, and Save as artifact, and answered clarification cards compact into per-question reviewable records. New model options include OpenCode Go and OpenCode Zen as built-in API-key providers and GLM-5.3-Flash for Zhipu AI. Under the hood, conversation rendering lazy-loads Mermaid and code-highlighting runtimes, long-running sessions persist at a bounded cadence instead of once per frame, archiving waits for active reviewer and remote compute work, and a set of session persistence, plan preview, and project-file fixes lands throughout.

✨ Highlights

• Spanish interface. The full interface — onboarding, settings, conversation surfaces, native dialogs, and release notes — is now available in Spanish, joining the existing seven languages with a runtime switcher in Settings. (#1771)
• In-app source previews. Links in agent responses open in a sandboxed preview inside the app: hovering reveals the source title and full URL, and a click loads the page in the side panel with a deterministic progress indicator, without leaving your workspace. (#1524)
• Live notebook variables. A new Variables view inspects the running Python or R namespace — names, types, shapes, and previews — read-only, refreshed after each execution, with no kernel started just to browse. (#1748)
• Session hover previews. Hovering or focusing a session in the sidebar shows its title and description, and overflowing titles scroll so long names stay distinguishable. (#1775)

🚀 New Features

• Spanish localization — complete common, native, and renderer catalogs with neutral international Spanish, native Electron messages, date formatting, and localized documentation. (#1771, #1780)
• In-app source previews — HTTPS links in agent responses become native source links with an interactive popover, in-panel sandboxed loading with a toolbar progress indicator, external-browser shortcut, keyboard navigation, and preserved URL display. (#1524)
• Live namespace browser — a second-level Variables view for notebook kernels with filtering, private-name toggle, manual refresh, and stale/refreshing/unavailable states; snapshots are bounded and never persisted. (#1748)
• Session hover previews — immediate title-and-description preview on hover or keyboard focus, reduced-motion support, and desktop-only gating. (#1775, #1796, #1797)
• Right-click preview-tab actions — Close, Close others, plus context-sensitive Download, Copy path, and Save as artifact, anchored at the pointer without activating the tab. (#1764)
• Elicitation cards with per-question review — answered and skipped question cards become compact records whose answers expand back to the original questions, with accurate selection tallies and compact controls. (#1772)
• New providers and models — OpenCode Go and OpenCode Zen as built-in API-key providers, plus GLM-5.3-Flash alongside GLM-4.5-Air and GLM-5.3 for Zhipu AI (GLM). (#1763, #1790, #1762, #1766)

🔧 Improvements

• Conversation rendering loads Mermaid and code-highlighting runtimes only when a message actually contains them, shortening renderer startup. (#1789)
• Long-running sessions persist at a bounded cadence instead of once per presentation frame, removing sustained CPU, memory, and disk pressure on large sessions. (#1779)
• The response footer labels its model-request summary as calls, consistent with the context-window view. (#1781)
• Archiving a project now waits for active reviewer work and non-terminal remote compute jobs, and pauses the queued message pipeline until the project is restored. (#1785)
• Long plan summaries are clamped to three lines with hover reveal, and the plan preview keeps its scroll position across streamed progress updates. (#1783)

🐛 Bug Fixes

• Sessions — permission approvals no longer collide with title/description generation, preserving both instead of surfacing a persistence alert. (#1768)
• Sessions — interrupted sessions resume with the authoritative failure preserved when providers report structured errors, instead of silently resetting context. (#1774)
• Projects — a configured project agent context is enforced consistently: lookup failures fail closed and context edits apply to idle sessions before the next prompt. (#1786)
• Project files — granted-folder permission changes that fail now surface a retryable explanation instead of silently keeping the old grant. (#1793)
• Notebook — local RPC requests are strictly validated per method, rejecting malformed parameters before execution. (#1794)
• Session previews — hover previews dismiss immediately and keep working after the pointer bridge changes. (#1796, #1797)

📦 Install

Requirements: macOS 12+ (Apple Silicon or Intel), Linux x64, or Windows 10/11 x64. On first run, the onboarding wizard checks the environment and can install and configure an app-managed agent runtime. Once installed, the app can update itself in place.

Download the appropriate package from the Assets section below:

Platform	Package
macOS (Apple Silicon)	DMG for ARM64
macOS (Intel)	DMG for x64
Linux	AppImage or Debian package for x64
Windows	Installer for x64

macOS — first launch. Official release builds are Developer ID signed and notarized by Apple, so they open like other trusted applications. A locally built copy is not notarized and may require approval through macOS Privacy & Security.

Windows — first launch (unsigned build). No Authenticode certificate yet, so SmartScreen shows a bypassable "unrecognized app" prompt (More info → Run anyway). Verify that the package came from the official release page before continuing.

Build from source instead:

npm install
npm run build:mac   # or: build:linux / build:win

🧭 What's in this release (maturity)

• ✅ Implemented: a local-first desktop, localhost-web, headless, CLI, and task-SDK surface over persistent projects and sessions with selectable message branches, branching into a new session from user messages or completed agent messages with persisted source lineage, composer session references (#) with turn-scoped read access, reversible archiving with keyboard undo, project pinning, collapsed long user messages, persistent side conversations with advisories injected into running main turns, generated and editable session details, session hover previews in the sidebar, and SQLite-indexed summary-first session startup; in-app sandboxed previews for source links in agent responses; selectable Claude Code, OpenCode, and Codex agent frameworks behind a shared provider turn-adapter interface; production subagent delegation with durable messaging, restart recovery, structured output, artifact and review evidence, and camelCase delegation APIs; review-gated session plans with CLI plan controls; a unified composer lane with a session-scoped message queue, unified draft undo and redo history, active-specialist emphasis, and mid-turn Send now through native follow-up steering; hot-switching of compatible models and providers; multi-provider model configuration including an xAI OAuth subscription, a dedicated Vision model selector with persistent evidence relay, custom token limits, and a consolidated Scenario models card; per-model-call usage details with a per-call context-window chart; a configurable reviewer model policy with an isolated review runtime, durable assessment snapshots, and empty initial checks; a token usage dashboard; context-window composition insights with compaction boundaries; persistent Python/R/REPL kernels with bounded run-history payloads, a read-only live variable browser for running kernels, separated approval and execution state, cross-run dependency tracking, package-installation progress, progressive history loading, terminated read-only previews, session-scoped remote SSH execution with key or password authentication, harvest ceasing, and remote file-staging cleanup; a user terminal shared with the agent; app-managed and bring-your-own environments for Python and R; immutable, session-scoped artifact versions with checksummed content, producer code, execution history, exact input references, environment inventory, producing message-branch context, version-scoped reviewer evidence, and on-demand LLM-driven code reconstruction; rich in-app previews for scientific data, documents, images, source code, molecular structures, and notebook history with right-click tab actions and full-screen mode; file attachments up to 10 GB with streaming upload; skills with conversational creation, import, marketplace browsing, and explicit / selection; 24 built-in research connectors plus custom MCP servers with import/export of standard MCP client configurations; durable scoped permissions with allow grants; remote-access pairing; interface localization in Spanish, French, Chinese (Simplified and Traditional), Japanese, Korean, and Russian; and auto-update with prominent update reminders and localized release notes.
• 🚧 Partial: R remains managed-only; provider choice remains constrained by the active framework's endpoint compatibility; remote compute is SSH-only; skills remain local (no hosted public discovery commons); and review is opt-in and record-scoped.
• 🗺️ Roadmap: a unified model gateway, a hosted public skills and specialist discovery commons, Slurm/cloud-GPU execution, stronger sandboxing and credential isolation, and collaborative research workflows.

🐢 Known Limitations

• R is managed-only. A bring-your-own R interpreter path is not built yet.
• Remote compute is SSH-only. Slurm and cloud GPU submission are not built yet.
• Provider choice is per framework, not one unified gateway. The available protocol depends on the selected agent backend.
• Hot-switching applies only to registered compatible targets. Framework, auth-lane, wire-route, or unsafe capability changes still require a reconnect.
• Code reconstruction is LLM-generated. It does not replace deterministic reproduction; portable environment locks and full-fidelity session replay remain open.
• No hosted public specialist discovery commons. Specialist packages are portable across machines via import/export and the signed marketplace; what is not built yet is a hosted public discovery and forking hub.
• The task SDK is a first-generation surface. Task creation, polling, artifact retrieval, run progress, and cancellation work; broader orchestration remains open.
• Switching agent backends cannot transfer in-flight tool state. Existing conversation history can replay, but a running action is not migrated.
• Skills are local only. There is no shared public commons, cross-machine forking, or user-facing version pinning yet.
• The reviewer is opt-in and record-scoped. It does not replace domain-specific validation of citations, units, statistics, or methods.
• Scoped permissions cover allow-grants only. Network sandboxing, directory-level file access control, and a credential vault are not built yet.
• Windows builds are unsigned. SmartScreen may warn on first launch; official macOS builds are notarized.
• No local GPU compute backend.
• No multi-user real-time collaboration.

🙏 Acknowledgements

Thanks to @ewen-poch, @wen2zhou, @daanveer-tech, @justemu, and everyone in Discord, X, and Discussions.

---

Full Changelog: https://github.com/aipoch/open-science/commits/v0.20.2