AIPOCH Open-Science v1.2.1

AIPOCH Open-Science v1.2.1

A unified Resource Center for skills, connectors, and specialists, project-scoped prompt optimization, Codex subscriptions and Claude Fable 5.2 in the built-in Agent, compact evidence for long research sessions, and more reliable Windows updates and cloud-compute sign-in.

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.

v1.2.1 follows v1.2.0 with a workflow-and-reliability release. Skills, Connectors, Specialists, and their Marketplace move into one unified Resource Center with shared cards, filters, and detail views, and Marketplace installation now works in the Web renderer too. The composer gains manual, project-scoped scientific prompt optimization with review, edit, and undo, plus custom templates in Project settings. The built-in Open-Science Agent can now use Codex subscription accounts for the main agent and subagents, adds Claude Fable 5.2 with a 1M context window, and no longer truncates provider-supported long answers at 4,096 tokens. Notebook and Compute runs store their evidence once in a compact, verifiable form, so .science exports of long research histories stay fast and small. Modal and Hugging Face compute targets provision a compatible Python runtime before browser authorization and verify accounts through the system proxy. Windows update packaging, recovery, and diagnostics improve, answered Agent questions no longer stall before continuing, research film generation survives session switches, and idle interpreters free their memory automatically.

✨ Highlights

• One place for every resource. The Resource Center now manages skills, connectors, and specialists — including Marketplace installation, also in the Web UI — with shared cards, filters, and detail views, and Skill references open in the project pane without leaving the conversation. (#4270)
• Scientific prompt optimization. Refine the current project's prompt from the composer with one click; review, edit, or undo the result, and save custom templates in Project settings. Optimization only runs through your configured provider when you invoke it. (#3991)
• A stronger built-in Agent. Codex subscription accounts are now available for the main agent and subagents, Claude Fable 5.2 joins the model list, and long answers are no longer cut short at 4,096 tokens. (#4250, #4249, #4275)
• Long research sessions export without stalling. Notebook and Compute runs store their evidence once in a compact, verifiable form, so .science exports of long histories stay fast and small. (#4248, #4258, #4240)
• Cloud compute sign-in that survives real networks. Modal and Hugging Face provision a compatible Python runtime before opening the browser, and account verification now follows your system proxy. (#4255)

🚀 New Features

• Manage Skills, Connectors, and Specialists — and install from the Marketplace, including in the Web renderer — from the unified Resource Center, with consistent catalog cards, filters with reset, bulk selection, and detail drawers; Skill references from the composer and sent messages open in the project Resource Center pane without disturbing the active conversation or draft. (#4270)
• Optimize the active project's prompt from the composer: review and edit the result before it replaces the draft, Undo restores the previous text and any removed pasted attachments, and custom optimization templates live in Project settings. Optimization is manual and runs through the configured provider only when invoked. (#3991)
• Use imported or app-owned Codex subscription accounts for the main agent and subagents in the built-in Open-Science Agent, with sign-in, refresh, and credential replacement reusing the existing Codex connection flows. (#4250)
• Select Claude Fable 5.2 with a 1M context window from the bundled Anthropic model list. (#4249)
• Export bibliography metadata as CSV alongside BibTeX and RIS — selected references across pages with individual exclusions, or an entire project or collection — using 16 stable English column names, with empty cells for missing values. (#4262)

🔧 Improvements

• The built-in Agent honors each model's advertised output limit (the configured value, then exact vendor metadata, then a 16,384-token fallback) and prefers each model's native endpoint protocol, so provider-supported long answers arrive complete. (#4275)
• The built-in Agent measures its context usage correctly and compacts much later in long sessions. (#4209)
• An answered Agent question no longer waits on a deferred persistence flush, removing the "completion has not committed yet" stall before the answer continues. (#4279)
• Modal and Hugging Face compute targets share one Python-runtime preparation flow: the selected interpreter is checked without installing into it, authorization and SDK calls stay bound to it, and a stale authorization result can no longer overwrite a newer connection. (#4255)
• Notebook and Compute evidence is stored once per unique content with verified digests; exports of long histories — hundreds of runs over the same files — produce compact packages, and the first export backfills historical evidence. (#4248, #4258, #4240, #4259)
• Research film generation continues when you switch sessions, instead of stopping with the conversation that started it. (#4268)
• Idle Python, R, REPL, and Shell interpreters are reclaimed after 15 minutes without session activity, with small per-kind pools, so finished conversations no longer accumulate resident runtimes. (#4274)
• The context-reset notice dismisses itself after six seconds, and resource catalog cards keep their usage avatars compact. (#4281, #4276)

🐛 Bug Fixes

• Notebook files attached in chat open a preview again, and local file links in replies open the corresponding file. (#4257)
• Signed Windows update packaging is unblocked, and update recovery names the files and processes holding the app, prunes closed startup logs, and resolves blockers without wedging the installer. (#4233, #4230, #4231, #4247)
• Literature imports preserve identifier routing and DOI letter case. (#4225)
• Website source previews are responsive again in narrow layouts. (#4220)
• External PDF structure recovery hardens against damaged files, and PDF translation isolates per-document agent prompts and tool capabilities. (#4238, #4228)
• Mobile pairing stays off until you select it, and web remote-control RPC failures explain what happened instead of failing silently. (#4221)
• Provider error details survive through the agent runtime, and OpenCode request-classification failures no longer leak beyond the request that hit them. (#4237, #4261)
• Conversation commits acknowledge terminal events reliably, authorized local CLI imports work again in session packages, and first-message navigation is restored in long conversations. (#4269, #4267, #4260)
• Linux recovery no longer mistakes confirmed kernel threads for leaked Notebook processes. (#4265)

📦 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. Linux packages additionally arrive as standalone .deb and .rpm installers.

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 v1.2.0, plus the unified Resource Center for skills, connectors, and specialists, project-scoped prompt optimization, Codex subscription accounts in the built-in Agent, Claude Fable 5.2, bibliographic CSV export, compact evidence storage for long Notebook and Compute histories, and the fixes described above.
• 🚧 Partial: remote compute covers direct SSH, Slurm, and user-connected Kubernetes, Hugging Face Jobs, Modal, and Ray targets, each requiring user-provisioned accounts, software, and permissions; provider choice remains constrained by the active framework's endpoint compatibility; 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; review is opt-in and record-scoped.
• 🗺️ Roadmap: a unified model gateway, cross-machine skill forking and version pinning, 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.
• Imported .science sessions stay read-only at the import itself. They remain inspectable and referenceable; step-linked discussions and forking provide writable continuations without modifying the import.
• 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.
• OCR recognition data downloads on demand. Scanned pages are recognized after the matching language pack is installed, and coverage indicators show which pages still lack text.
• AI study snapshots reflect the configured main model. They summarize the available text at generation time and are not a substitute for reading the paper.
• Research film narration needs a locally installed voice model. Generation offers to install the narration voice on first use; films are produced once the voice model is ready. The voice downloads from the app's built-in CDN first and falls back to pinned upstream endpoints if the CDN is unreachable; downloads are verified against the published catalog.
• 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.
• WSL2 Bash is a preview. It is opt-in on Windows x64 only; PowerShell remains the default shell runtime.
• Remote compute requires user-provisioned targets. Direct SSH, Slurm, and the Kubernetes, Hugging Face Jobs, Modal, and Ray backends run under the approved target's account and need the host, software, resources, and permissions described in the remote-compute FAQ; jobs on cloud backends may incur costs on the connected account.
• 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.
• Mobile pairing over the internet routes through Pinggy's servers. Local WiFi pairing stays on your network; the internet tunnel mode discloses the relay in the connection dialog, and pairing stays off until you enable it.

🙏 Acknowledgements

Thanks to @ewen-poch, @daanveer-tech, @wen2zhou, @nasus2002, @roxi3906, @guch96, @cyj1116 and everyone in Discord, X, and Discussions.

---

Full Changelog: https://github.com/aipoch/open-science/commits/v1.2.1