AIPOCH Open-Science v0.37.1

AIPOCH Open-Science v0.37.1

A fidelity-and-hardening patch: scientific PDF structure extraction keeps complete figures and ruled tables, the Windows experience handles non-ASCII installation paths and quiet Notebook subprocesses, and installers shrink back down after the v0.37.0 runtime split.

AIPOCH Open-Science is an open-source, local-first, model-agnostic 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.

v0.37.1 follows the v0.37.0 runtime re-architecture with a focused set of fixes. PDF structure extraction again preserves what the source document owns: figures keep their captions and legends instead of splitting into partial crops, ruled tables next to formulas survive reconstruction with their grouped headers and row structure intact, native extraction no longer assigns captions, graphics, or cell text to neighboring elements, and translated documents regenerate reliably even when embedded fonts expose the same glyph under different names. On Windows, the desktop CLI now works from installation paths containing Chinese and other non-ASCII characters, local PDF translation handles Windows drive paths, and Notebook work no longer flashes console windows while kernels and environment captures run. Installers also drop the unused runtime payloads that inflated v0.37.0, restoring leaner downloads.

✨ Highlights

• PDF fidelity restored. Structure extraction keeps complete figures with their captions and legends, reconstructs ruled tables that sit next to formulas without rejecting them, and preserves native ownership of figure regions, table records, and source-proven text. (#3456, #3465)
• A quieter, more robust Windows experience. The desktop CLI works from non-ASCII installation paths, PDF translation handles Windows drive paths, and Notebook kernels and environment captures no longer open console windows. (#3457, #3460)

🔧 Improvements

• Desktop installers and the standalone runtime shed unused payloads — duplicate bundled dependencies, unused model-runtime and tokenizer assets, and platform libraries the app never loads — reversing the size growth introduced by the v0.37.0 backend split (#3451, #3470)

🐛 Bug Fixes

• A spaced panel reference in running prose no longer splits a figure into separate raster and vector crops, and detector padding below a table's closing rule no longer causes the whole table to be rejected as a damaged formula (#3465)
• Native PDF extraction keeps caption and graphics ownership with the right element, recovers independently bounded table records with grouped headers and row groups, and retains cell source references and rich text (#3456)
• The desktop CLI no longer misreads its launcher under legacy code pages when the installation path contains Chinese characters, and preserves the caller's code page and exit code (#3457)
• Local PDF translation on Windows no longer fails with an unsupported module URL when the WASM assets live on a drive path (#3457)
• Notebook kernels, environment fingerprinting, package inventory, and lock capture no longer flash a console window or Windows Terminal during execution (#3460)
• Translated PDF regeneration no longer rejects legitimate source text because of differing embedded glyph-name spellings — proven glyphs keep their original ink (#3469)

📦 Install

Desktop client. Requirements: macOS 12+ (Apple Silicon or Intel), Linux x64 or ARM64, 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 updates itself in place.

Download the appropriate package from the Assets section below, or from the official download page. On macOS, the desktop app is also installable via Homebrew: brew install --cask open-science.

Platform	Package
macOS (Apple Silicon)	DMG for ARM64
macOS (Intel)	DMG for x64
Linux	AppImage or Debian package for x64 / ARM64
Windows	Code-signed installer for x64

Command line via Homebrew (macOS and Linux). The standalone CLI and local Web UI live in the official tap — bundled Node runtime, no separate Node.js needed, and upgrades through the normal brew flow:

brew install aipoch/tap/open-science
open-science start --no-open
open-science url

Command line via npm. With Node.js 22.13+, install the CLI (and Task SDK) globally; updates then follow your normal npm upgrade flow:

npm install --global @aipoch/open-science
open-science --help

Command line standalone (all platforms, no toolchain). The Assets below also include standalone CLI archives — macOS and Linux arm64/x64 tarballs and a Windows x64 zip — with a bundled Node runtime, so they install and run without Homebrew, npm, or a local Node.js.

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 notarized differently and may require approval through macOS Privacy & Security.

Windows — first launch. Official release builds are code-signed; Windows recognizes the publisher and no SmartScreen warning appears. 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: everything shipping in v0.37.0, plus the PDF structure fidelity fixes, Windows path and console fixes, and installer slimming described above.
• 🚧 Partial: provider choice remains constrained by the active framework's endpoint compatibility; remote compute covers direct SSH and Slurm (cloud-GPU submission is not built yet); WSL2 Bash remains an opt-in Windows x64 preview; the skills commons ships marketplace discovery but not cross-machine forking or user-facing version pinning; reproducibility ships replayable per-version verification, session replay, and RO-Crate packaging, while solver-exact equivalence remains open; conditional package restore is an exported script workflow; and review is opt-in and record-scoped.
• 🗺️ Roadmap: a unified model gateway, cross-machine skill forking and version pinning, cloud-GPU execution, stronger sandboxing and credential isolation, and collaborative research workflows.

🐢 Known Limitations

• Replay reproduces recorded steps, it does not re-execute them. Solver-exact equivalence guarantees remain open; replayable verification remains the deterministic check, and starting a check from the web client currently requires the desktop app.
• Sensitive package evidence is opt-in. Session diagnostics include sensitive package evidence only after explicit confirmation, and the export stays local until you share it.
• Imported .science sessions are read-only. They stay inspectable and referenceable, and forking them into a writable continuation remains available; execution and continuation remain disabled on the import itself.
• Bookmarks belong to a single session. They do not follow branch switches, join .science package exports, or sync across machines — forking copies them into the new session with fresh identities.
• PDF annotations are bound to the file version they annotate. Notes follow the file across projects and sessions where that version is shared, but they do not sync across machines.
• WSL2 Bash is a preview. It is opt-in on Windows x64 only; PowerShell remains the default shell runtime.
• Remote compute covers direct SSH and Slurm. Cloud-GPU submission is not built yet.
• Editable artifacts cover text formats. Binary formats stay read-only, and editing always publishes new versions.
• Network sandboxing covers the app's notebook and compute runtimes, not the whole system. On Windows, the boundary applies only after the sandbox's one-time administrator setup.
• Provider choice is per framework, not one unified gateway. The available protocol depends on the selected agent backend, and remote endpoints must use HTTPS.
• Code reconstruction is LLM-generated. It does not replace deterministic reproduction; replayable verification is the deterministic check.
• The reviewer is opt-in and record-scoped. It does not replace domain-specific validation of citations, units, statistics, or methods.
• No local GPU compute backend.
• No multi-user real-time collaboration.

🙏 Acknowledgements

Thanks to @ewen-poch and everyone in Discord, X, and Discussions.

---

Full Changelog: https://github.com/aipoch/open-science/commits/v0.37.1