Wisp Science releases

Current release

v1.8.0

Archived release

v1.7.1

Archived release

v1.7.0

Archived release

v1.6.0

Archived release

v1.5.0

v1.5.0 is the Share release. Conversations can leave the app as a long PNG or a self-contained HTML page that matches the live chat, from /share, the topbar, or the composer menu. It also adds Grok Imagine image and video generation, isolates Python/R runtimes per conversation, and tightens session, browser, and upgrade reliability.

Archived release

v1.4.0

v1.4.0 turns remote servers into disposable compute rather than permanent file centers. It closes the SSH task loop from upload and background execution through output harvesting, review, transfer recovery, and safe cleanup, while adding server storage preferences, remote-file ownership, scheduled automations, and a broad set of workflow and UI refinements.

Archived release

v1.3.0

v1.3.0 focuses on long-running conversations and the model experience: it adds real long-context memory evaluations, a dynamic context-usage gauge, and per-model reasoning-effort controls. It also improves slash commands, global habits, and end-of-turn handling, while fixing streaming responses, repeated automatic compaction, MCP child-process shutdown, and several UI issues.

Archived release

v1.2.0

v1.2.0 ships official Linux packages for the first time (deb / AppImage for x86_64 and aarch64), restores Exploration branches as mutually isolated workflows with several state-leak fixes, and brings query-aware project memory retrieval, per-session hot-reload of project rules, plus a round of plugin, ACP, and UI stability fixes.

Archived release

v1.1.0

Archived release

v1.0.0

v1.0.0 是 wisp-science 的首个正式版本：移除 Beta 标识，并带来探索分支（Exploration Branches）这一核心能力——可以从任意历史轮次创建隔离分支、安全地提升或丢弃，让科学分析中的"试错"不再污染主线会话。本版本还包含一轮全面的界面打磨、前端性能优化与多项稳定性修复。

Archived release

v0.36.0

v0.36.0 聚焦长会话的可发现性与响应速度：新增侧边栏会话搜索和空会话即时展示，显著降低大型工具会话回放与继续对话时的卡顿，并让停止任务、模型等待与工作流模型选择更加清晰可靠。

Archived release

v0.35.0

Archived release

v0.34.0

Archived release

v0.33.0

v0.33.0 聚焦长时间运行任务与上下文管理：本地和 WSL 的 Run 现在以分离（detached）方式执行并移除了 300 秒超时上限；上下文压缩检查点经过重新设计，配合子代理轨迹、溢出恢复等改进，让长会话更稳定。同时新增将 Skill 一键转换为 Workflow 的 create_workflow 工具，并带来多项界面与安装体验改进。

Archived release

v0.32.0

wisp-science v0.32.0 strengthens Agent Workflows for long-running research. Workflow-native method search can reuse recorded methods and activities, self-explanation makes plans and results easier to inspect, and failed plans can resume with adjusted budgets. Cross-conversation resource leases, bulk conversation actions, and automatic context-compaction recovery make parallel work more reliable.

Archived release

v0.31.0

wisp-science v0.31.0 makes Skills auditable and planable research capabilities. Semantic YAML manifests describe their meaning and scope, Catalog Snapshots preserve the available inventory, and budget-aware Portfolios bind exact Skills to workflow nodes. The release also adds editable project Memory, environment reference cards, a research-design template, and stronger Agent pause, question, and context-transfer safety boundaries.

Archived release

v0.30.0

wisp-science v0.30.0 introduces Publication Evidence Capsules: project-level, auditable packages that connect immutable Artifact versions, Runs, source evidence, readiness manifests, and isolated verification. The new Publication workspace lets researchers select evidence deliberately, assess readiness, freeze a snapshot, and verify fine-grained claims without relying on mutable local state.

Archived release

v0.29.0

wisp-science v0.29.0 brings multi-Agent orchestration to the desktop through Quick Action and Workflow Studio. Create, arrange, and connect workflow nodes visually, then follow real execution activity in the streamlined Agents panel. The release also adds project-free scratch chat, a roundtable workflow template, one-shot JSONL CLI runs, and an isolated Agent evaluation suite.

Archived release

v0.28.0

wisp-science v0.28.0 introduces optional, signed in-app updates for macOS. Users review release notes, explicitly download, verify the package signature, and confirm installation and restart. The updater selects the matching Apple Silicon or Intel artifact and refuses installation while active Agent work, approvals, reviews, or persisted jobs exist. This release also improves plugin install/update flow, preserves cross-project session-transfer selection, fixes image-model validation fallback, an

Archived release

v0.27.0

wisp-science v0.27.0 brings reviewable Plan mode and blocking user questions to built-in Agents, while improving device integration and everyday code reading. Structured plan cards support approval, changes, and save-and-exit actions; planning limits execution to read-only retrieval tools. The release also adds unified ask_user question cards across built-in and ACP paths, experimental StickS3 device-bridge endpoints, and one-click code-block copy.

Archived release

v0.26.0

wisp-science v0.26.0 strengthens traceable research conversations and day-to-day Agent collaboration. It adds safe Undo for the latest completed turn, nests branch sessions beneath their source, surfaces workspace Artifacts and metadata in the research graph, and lets users update or delete user Skills. ACP Plan mode controls, cursor-aware references, selection handoff, run monitoring, multi-window notification routing, and Escape handling are also more robust.

Archived release

v0.25.0

wisp-science v0.25.0 improves creation, reading, and project navigation. The new Scientific Illustrator can use a configured gpt-image-2 profile to generate PNG scientific images, while SVG/vector/editable requests retain the inspected SVG workflow. Image previews support numbered region comments and a one-click, review-before-send revision request. The research graph is now a project-level modal with list and graph views, and onboarding, model capability validation, browser-tab cleanup, convers

Archived release

v0.24.0

wisp-science v0.24.0 brings key research-workbench records into the visible UI and strengthens everyday reliability. The real-browser bridge can now return the visible viewport as an image for vision-capable agents; the right pane exposes the project's research graph and database-registered Artifacts; and Run cards link directly to their outputs. This release also makes directed SSH trust edges visible and revocable, adds workspace Storage and per-session Usage views, and improves long chats, li

Archived release

v0.23.0

wisp-science v0.23.0 makes it much easier to continue external Agent work and research code in Wisp. Import Codex CLI and Claude Code conversations from local, WSL, or configured SSH contexts with previews, caching, and idempotent incremental updates. Starred Library code now has immutable version history, Artifact provenance can be edited into a reviewed rerun request, and long conversations have a navigable outline.

Archived release

v0.22.0

wisp-science v0.22.0 adds structured, auditable transfers across SSH contexts and from SSH to local storage; editable queued follow-ups and pinned conversations; reusable dynamic Agent workflow drafts; and clearer MCP, Skill, and provenance surfaces. Remote command failures now remain correctable exploration instead of being confused with authentication failures, while cut model streams surface an explicit incomplete-response error.

Archived release

v0.21.0

wisp-science v0.21.0 adds installable feature plugins that bundle Skills, local MCP servers, and sandboxed MCP Apps; a persistent bridge to the user's existing Chrome/Chromium profile; and clearer multi-project workflows. Motif for Claude Science is the first end-to-end compatibility target, including guided Skill startup, an interactive workbench, artifact generation, and native Windows Node entry paths.

Archived release

v0.20.0

wisp-science v0.20.0 puts context management under explicit user control. Automatic history rewriting is gone: Wisp warns near the context limit, while /compact archives the full conversation before reducing model context. This release also adds a read-only explore subagent, migrates bundled scientific skills to Wisp's native Run/tool contract, separates progress, reasoning, and tool activity in chat, and improves per-reply token accounting and model settings.

Archived release

v0.19.0

wisp-science v0.19.0 replaces fixed Agent teams with dynamic, reviewable, persisted delegation. The main Agent can create capability-scoped Native or ACP workers, run dependency-aware tasks inline or in the background, delegate one bounded nested level, and isolate parallel writers in Git worktrees. Child conversations and model choices are now scoped to their owning conversation.

Archived release

v0.18.0

wisp-science v0.18.0 focuses on context transparency and session organization: export the exact request sent to the model with per-section token breakdown, progressively disclose MCP tools and skills, add a project-wide session reader, sort/group the session sidebar (folders renamed to groups), persist long-run progress, customize send/newline shortcuts, and surface an in-app update prompt with changelog. Fixes ACP agent delete confirmation, Escape scoping in the confirm dialog, and side-chat au

Archived release

v0.17.1

wisp-science v0.17.1 gives Manual, Assisted, and Automatic Agent delegation distinct real semantics, adds risk-gated automatic execution, and fixes recent panel, scrolling, clipping, and draft-control regressions.

Archived release

v0.17.0

wisp-science v0.17.0 adds controlled, persisted Agent delegation workflows with explicit approval and bounded permissions; fully offline XLSX and PPTX previews; saved chat highlights; background task and approval notifications; and quick model setup with reasoning-effort controls.

Archived release

v0.16.2

wisp-science v0.16.2 reuses one persistent SSH session per host for environment probes, Run polling, and remote file RPCs, preventing repeated TCP/authentication handshakes from looking like brute-force traffic. It also separates a host's display name from its real HostName and adds a connection test to the host form.

Archived release

v0.16.1

wisp-science v0.16.1 adds keyring-backed SSH password authentication, requires a successful probe before an SSH host can be used by the Agent, stops repeated managed login attempts after connectivity/authentication failures, and provides actionable host-edit and re-probe guidance. It also adds custom keyring-backed credentials exposed to new local Python/MCP processes, immutable DOCX/BibTeX message-resource previews, and consistent tab styling across the workspace.

Archived release

v0.16.0

wisp-science v0.16.0 introduces an AI-first script workflow with persistent Python/R runtime bindings, selected-code execution, live variables, a bounded console, and conversation-driven source edits. It also adds side-by-side document chat, richer PDF/DOCX/Markdown/notebook previews, confirmed image-region attachments, common local Files operations, SSH file previews, safer manual retry behavior for SSH failures, and direct pet navigation to sessions that need attention.

Archived release

v0.15.0

wisp-science v0.15.0 adds Feishu/Lark and WeChat remote conversations, session-scoped SSH/WSL compute authorization, a Python/R runtime environment explorer, OAuth for remote MCP connections including optional Notion setup, configurable agent iteration limits, single-instance desktop activation, and broad remote-access, settings, OAuth, layout, and Windows troubleshooting improvements.