AIPOCH Open-Science v0.37.0

AIPOCH Open-Science v0.37.0

An architecture-and-reach release: the shared backend runs as a standalone Node service, the connector catalog grows ENCORI plus deeper ChEMBL and ClinVar evidence, and scientific PDF translation and structure extraction preserve native fidelity end to end.

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.

v0.37.0 re-architects the runtime and widens scientific reach. The shared services behind the desktop, web interface, CLI and Notebook now run in an ordinary Node process: open-science start no longer needs a windowless Electron host, the desktop app connects to the backend as a native client, and independent desktop backends automatically allocate free ports instead of colliding. The connector catalog grows a built-in ENCORI connector with ten tools for miRNA targets, RNA–RNA interactions, regulatory evidence, reference tables and bulk datasets; ChEMBL adds assay details and bioactivity pagination beyond the first 1,000 records; and ClinVar adds individual submission evidence for comparing submitters' classifications. PDF translation preserves native visual reading order and verifies nested ActualText labels, while structure extraction keeps source-owned figures and complete table records. Notebook runtimes report mixed Conda/pip environments accurately, preserve micromamba build metadata, and present streamlined approval cards for ready default runtimes. Fixes cover Windows Python discovery, desktop backend profiles, .science imports into new projects, smart-collection batch retries, and Codex streamed tool arguments.

✨ Highlights

• A standalone Node backend. The shared services behind the desktop, web, CLI and Notebook now run in an ordinary Node process — the command line and headless mode no longer need a windowless Electron host, the desktop app connects as a native client, and desktop backends automatically pick a free port. (#3358, #3393)
• Connector reach. A new built-in ENCORI connector brings miRNA–target and RNA–RNA interaction evidence, ChEMBL adds assay details and bioactivity pagination, and ClinVar adds individual submission evidence. (#3376, #3343, #3361)
• Fidelity for scientific PDFs. PDF translation preserves native reading order, nested ActualText labels verify correctly, and structure extraction keeps source-owned figures and complete table records. (#3360, #3348, #3370)
• Calmer Notebook approvals and capture. Ready default runtimes no longer double-prompt, mixed environments report accurately, and package build metadata stays visible. (#3380, #3398, #3385)

🚀 New Features

• Standalone Node backend for shared services — open-science start runs without Electron, and the desktop app connects as a native client (#3358)
• ENCORI connector with ten tools for miRNA targets, RNA–RNA interactions, regulatory evidence, reference tables and bulk datasets (#3376)
• ChEMBL connector gains assay details and bioactivity pagination beyond the first 1,000 records (#3343)
• ClinVar connector gains individual submission evidence for comparing submitters' classifications (#3361)
• Home empty state with guidance and a Create project action (#3371)

🔧 Improvements

• Notebook environments: accurate mixed Conda/pip capture, preserved micromamba build metadata in the package table, and hardened callback dependencies and file-input lineage (#3398, #3385, #3369)
• Linux credentials stop probing the Secret Service backend after the first secret operation, preventing spurious fatal recovery exits (#3389)
• Session plan approval and dismissal no longer interfere with execution recovery (#3366)
• GitHub releases now ship standalone CLI archives (macOS and Linux arm64/x64, Windows x64) alongside the desktop installers, so the CLI no longer requires a local Node.js toolchain (#3394)

🐛 Bug Fixes

• Desktop backends allocate a free port, fixing startup failures when another backend is already running (#3393)
• Windows Python discovery no longer misinterprets interpreter paths with spaces or parentheses (#3362)
• The desktop runtime preserves the selected backend profile on Windows (#3395)
• Importing a .science package into a new project no longer fails during catalog adoption (#3379)
• Smart collections resume and retry the failed batch instead of starting over (#3351)
• Codex recovers streamed MCP tool arguments that arrive empty in the completed call (#3349)
• Switching Specialists on Windows no longer leaves an approved handoff pending and blocking later prompts (#3388)
• The PDB connector rejects malformed search responses instead of reporting zero hits, and incomplete pagination is no longer marked as finished (#3399)
• Auto mode stops asking for redundant tool approvals when inspecting or cancelling a background run, listing environments, reviewing a plan, or previewing a skill import (#3408)
• Notebook cells that fail before any source executes no longer linger in same-kernel risk history and force repeated approvals on follow-up cells (#3410)
• Headless hosts and SDK consumers again resolve the notebook runtime's bundled resources — execution loops, evidence capture, and source-analysis grammars (#3406)

📦 Install

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 can update itself in place.

Download the appropriate package from the Assets section below, or from the official download page:

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. Releases also include standalone CLI archives — macOS and Linux arm64/x64 tarballs and a Windows x64 zip — that install without a local Node.js toolchain.

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 v0.36.0, plus: the shared backend running as a standalone Node service with the desktop as a native client; the ENCORI connector; ChEMBL assay details and bioactivity pagination; ClinVar submission evidence; the Home empty state; and PDF translation/structure fidelity improvements.
• 🚧 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, @nasus2002, @wen2zhou, and everyone in Discord, X, and Discussions.

---

Full Changelog: https://github.com/aipoch/open-science/commits/v0.37.0