Features
Every feature, with its limits
What Conduit does today, area by area — and, next to each item, the default or the limit that applies to it. Nothing below is on a roadmap; the last section lists what is deliberately not built.
Personal apps
Ask for a tool and get a working page — a dashboard, a converter, a tracker, a game — then keep it. See personal apps in the docs, or the product tour for a screenshot.
- Save a page as an app. Save as app in a page's ⋯ menu makes a copy you open from Apps on the rail, and it keeps working after you delete the chat. Saving asks which of the page's site permissions to keep, one by one, and keeps none you don't tick.
- Eight starter apps built in. A Pomodoro timer, unit converter, Snake, a memory game, a world-capitals quiz, a budget tracker, and a weather dashboard and currency converter that use live data and ask before they connect.
- Data that stays between launches. A page or app can keep up to 5 MB of its own data, stored with the rest of yours and encrypted only when encryption at rest is on. Clear data removes it.
- Settings and your model. A page can declare a few inputs, such as a city, and Conduit draws the form. After you allow it for that page, it can also ask your AI model: the prompt names the provider and says when text would leave your device, and the page gets text back — no tools, chat history, memory, or documents.
- Per-app controls. Every personal app has its own Settings page: which model answers it, a daily token limit for cloud models (100,000 by default; models on your computer are never limited), its saved data with Export and Clear, the sites and providers it may use, and a week of activity records that never include prompts, replies, or saved values.
- Known issue on macOS. A page's script can still open a WebRTC connection there; Windows and Linux block it in the webview. Open pages and apps you got from someone else with care on macOS.
Slides
Presentations you build by asking. Describe the story, or ask for a deck in any chat. More on the Slides page and in the Slides docs.
- One live deck, edited in place. The assistant drafts a storyline for you to approve, then builds the slides into a single deck — each change edits that deck rather than producing another copy. The deck opens in a studio with the slide large in the middle and the chat beside it.
- Targeted edits. Ask for a change and the assistant edits one slide, or swaps a word across the whole deck.
- Built from layout parts. The assistant picks a layout and fills in its parts — headline, points, numbers with labels, columns, a quote — and Conduit lays the slide out in the theme. Text too long for its place is sent back to be shortened before the slide is made. A free-form layout remains for slides no layout fits.
- Charts from your numbers. Bar, line and funnel charts are drawn by Conduit from the data, with axes that start at zero and labels that don't collide; a full-width layout gives a chart the whole slide.
- Text that fits. Text that doesn't fit shrinks a step at a time, down to a readable size, instead of overlapping or running off the slide. Exported HTML, PDF and PowerPoint files match what you see.
- A layout check after each change. If text runs off, overlaps, crowds the footnote, or a diagram is unreadable, the assistant gets one round to fix it, and anything left is shown to you as a note. Your own text is never changed by it. It catches those problems; it doesn't guarantee good design.
- Script. Every word of the deck as one document. Type there, or double-click text on a slide; text you write is kept when the assistant rewrites a slide, unless you ask it to change that text.
- History. A version is kept for every change, labelled by your request, and you can restore any of them.
- Two built-in themes, plus any theme the assistant designs for you.
- Present with a presenter view. Present full screen, with a separate presenter window — current and next slide, speaker notes, and a timer — to keep private while you share the slides.
- Export. As a single HTML file that presents itself in any browser, or as a PowerPoint file (
.pptx), both on every platform, or as a PDF on Windows only. The PowerPoint file matches the deck as Conduit draws it: same fonts, colours, cards and positions. Text stays editable, charts are native PowerPoint charts that open with their numbers in Edit Data, and speaker notes come along. - One thing to know. Slides adds to your local database, so going back to a release from before Slides shipped starts with an empty database (the old file is kept as a backup).
Spreadsheets
Ask for a budget, a price list or a calculator and a spreadsheet opens beside the chat. See spreadsheets in the docs.
- Formulas that recalculate. The assistant writes formulas and Conduit computes them, with number, currency, percent and date formats, several sheets and a frozen header row. You change the sheet by asking, and the currency you ask for is the one used.
- Bar, line and pie charts. Ask for a chart and it appears under the sheet, in the app's colours for both themes. Charts follow the data when rows are added or removed.
- Excel in and out. Save as
.xlsx, with real Excel charts, which opens in Excel or Google Sheets. Attach an.xlsxfile to a chat, or add one to Documents, and the model reads its sheets as tables. - Rename by asking. The assistant can change a spreadsheet's title, for example after removing a product from “6 Products”.
Research
A research report whose every finding is checked against its source. More on the Research page and in the Research docs.
- A brief you approve first. Turn on Research in the composer's + menu, or pick it on Home. It proposes the sub-questions it will answer, a scope and a depth, which you edit and approve before anything runs. Quick is 8 searches, 15 pages and 10 minutes; Standard is 20, 40 and 20; Deep is 50, 100 and 40.
- Every finding checked against its page. It reads pages in full, web pages and PDFs alike, and pulls out facts with the exact words that support them. A fact counts only if those words are really on the page; the app checks that, not the model. It searches again for whatever is still unanswered.
- Sources rated for credibility. A weak source is marked in the report's sources, and its claims are attributed to it or left out, never stated as fact. Official, public-body and university sites are never rated low. The rating is the model's judgement; the quote check is the app's.
- A review pass, and today's date. Every sentence of the report is checked against the facts it cites, and what they don't support is corrected or removed. Research knows today's date, explains figures that differ by year, edition, definition or units, and names the sites it couldn't read when a sub-question goes unanswered.
- A report you keep. The result is a document in your chat: a summary, findings per sub-question, open questions, and numbered sources, with every finding cited. The chat card shows how many sources were cited and read. Stop keeps what was found so far.
- Pages can't give it orders. Page text only ever reaches a model call that has no tools, so a page that tries to give instructions can't make it search, fetch or save anything.
- Limits. It needs web search turned on and isn't available in local-only mode. A long page or PDF is read as an excerpt, the opening plus the parts that match the questions. A scanned PDF has no text to read. Your own Documents are not a source yet. A thinking model is slow, and Deep can hit its time limit.
Writing
Long pieces you plan as an outline and write with the assistant: blog posts, technical docs, reports, newsletters, essays. See the Writing docs.
- An outline you approve first. Describe what you're writing and the assistant proposes the sections, what each must say, and a target length. You edit and approve it, then the draft is written section by section into a Markdown editor next to the chat, until every section has text.
- Sourced facts, no made-up ones. A draft can look things up on the web (off by default), draw on your Documents collections, or work from a finished Research report with Write from this report. Facts from a source are linked, and the assistant is told to mark anything without one
[TODO]instead of making it up. - Your text stays yours. Any paragraph you type or change is kept word for word unless you ask the assistant to change it; Let AI edit hands it back. AI-written text can be shown tinted.
- Edit by selection. Select text for Rewrite, Shorter, Longer, Clearer, Fix grammar, or your own instruction; ask in the chat and only that part is rewritten.
- History and export. A version is kept after every change the assistant makes and after your own editing, and you can restore any of them. Export as Markdown, HTML or a Word file (
.docx) with real headings, lists, tables, links and code blocks. - Limits. Web lookups need web search turned on and aren't available in local-only mode. Only a finished Research report can be a source. Unlike Research, the app doesn't check a draft's facts against their sources; the links and
[TODO]marks are the model's work. Export is Markdown, HTML or Word only.
Workflows
Routines that run for you, when something arrives, and stop to ask. See the Workflows docs.
- Steps you can read. Fetch pages, search the web, run Research, search your Documents, have the model summarise, ask you a question, and save the result as a document. Start from a ready-made workflow — a morning briefing, a page watch, a weekly research digest, a weekly numbers deck — or build your own in the step editor, with problems shown before you save.
- Describe it, or save a chat. Say what you want in your own words — on the Workflows page, or with the Automate chip on Home — and the chat model drafts a workflow, checks it and fixes what doesn't check out. It opens unsaved, with a short list of what's left for you to pick; nothing is saved or scheduled until you say so. “Save chat as workflow” in the command palette turns what you did in a chat into one you can run again; it sees your messages and which tools were used, not the pages or documents themselves. Drafts can use branches, side-by-side steps, your connectors and helpers; anything the model invents, such as a connector you don't have, is removed and noted.
- A model per workflow, and per step. A fast, cheap model for the bulk of the work and a stronger one for the write-up, or a local model for steps that touch private data. Each run shows the model every step used. A workflow with no model chosen follows the chat model, and in local-only mode a step set to a cloud provider stops with a plain message.
- Run now, on a schedule, or when something arrives. Daily, on weekdays at a set time, or every few hours, with a notification when it finishes. With Run automatically on, a new post in an RSS or Atom feed, or a new file in a folder, starts its own run; the first check only notes what is already there. A workflow can also start when a connector's read-only list tool returns something new, checked every 15 minutes to once a day. A run missed while Conduit was closed happens once when you open it.
- Speaks up only when something changed. A “Continue only if…” step ends the run quietly when a value hasn't changed, is empty, or doesn't contain some text, and notify and save steps can skip themselves when the result is the same as last time. A scheduled run that finds nothing new sends no notification and shows “Nothing new” in the run list.
- Works on your files and your work. A workflow can have its own folder; “Read a file” reads text, CSV, JSON, Markdown, PDF or Word files from it and nowhere else, and “Turn text into a table” gives later steps rows to use or loop over. “Update a deck” and “Update a draft” change a saved Slides deck or Writing draft — text you wrote yourself is never changed, and a version is kept before and after so “Undo this update” puts it back. “Export a file” saves Markdown, text, CSV, JSON or HTML to the exports folder, and “Save to memory” only suggests a memory for you to accept.
- Connectors, with a say in what they change. A “Use a connector tool” step can list a folder, read a file or list a repository's open issues, and later steps can summarise what comes back. Tools that only read run on their own. A tool that sends, posts or edits something pauses the run and shows the exact call: Send, Edit it first, or Don't send. You can allow an action from now on in that workflow, or switch asking off under “Outside actions”; tools a connector marks as sensitive always ask.
- Asks before acting. Turning on automatic runs first lists everything the workflow will be allowed to do on its own — the sites it reads, web search, the folders it reads, connectors, which model and provider, saving documents — for you to approve. A run that needs more, after an edit or a settings change, pauses and asks: allow once, always allow, or don't allow. Research and Documents search steps ask before researching the web or searching your collections.
- Tools, with limits. A “Let the model use tools” step can search the web, read pages, check the time, or do arithmetic, and a scheduled run asks you to approve those tools first. The summarise step gets no tools. Each run has a time and token limit.
- See it as a canvas. An open workflow shows its steps as a canvas, or as a list if you prefer, and Conduit remembers the choice. The canvas draws the start, each step, loops and the lines between them, and during a run each step shows whether it is running, finished, failed, skipped or waiting for you. Select a step to see its input and output from a run. Add steps from a palette, drag them to reorder or into a loop, and drag a wire from one step to another to pick which output goes into which field.
- Branches and steps side by side. An “If… otherwise” step sends the run down a Yes or No path, and the path not taken shows as “Not taken”. A step can wait for any earlier steps, so steps that don't depend on each other run side by side, up to four at once, and a later step can join them. Workflows built before this run as they did.
- Results you can find. Every finished run keeps what it saved, with a preview. A workflow opens on its latest result, a Results tab lists them per workflow, and a Results list across all workflows marks the ones you haven't seen. Home shows “Fresh from your workflows”, and a scheduled run's notification offers Open.
- Agents that use tools and hand work to helpers. An agent step can use up to ten connector tools. A workflow can define helpers, each with its own instructions, model, tools and token limit, and an agent step hands tasks to them and gets their answers back. A helper can hand work on one more level at most, and a step hands out at most twelve tasks. The Agents page shows each agent and helper, whether it is working, and its tasks, failures and tokens over the last 30 days.
- Agents that use your computer, after you approve. This is off until you turn it on in Settings → Computer. A workflow's agent step or helper can then run commands and read, write or edit files in one folder you choose. Every command asks first unless a rule you allowed covers it, and some always ask, such as deleting files,
git reset --hardand force pushes. Once an agent has read outside content, such as a web page, allow rules stop applying for the rest of that run. File changes show a diff and ask first. A Permissions tab on each workflow lists its rules, how often each was used, and rules suggested from commands you approved in recent runs. On Windows, commands are also kept from writing outside the folder. macOS and Linux have no such confinement, so your approvals are the control. - Agents that browse, after you approve. With Settings → Computer on, a workflow agent or helper can open pages in your installed Edge, Chrome, Brave or Chromium, in a separate profile of its own, and read pages that build themselves with JavaScript. It can also click, type and choose if the workflow lets it. It asks the first time on each new site, and before anything that spends money, changes an account or signs in. The model never sees or types a password: you sign in yourself in the browser window, or save a site's sign-in in Settings → Computer, and after you approve, the app fills it in. You finish any human check yourself. Local and private-network addresses are refused. A download asks first, and an allowed file is kept with the conversation; programs, scripts and files over 100 MB are always refused. Site permission prompts (location, camera, notifications) are refused, and some sites refuse automated browsers; Conduit does not try to get around them. A Browser button beside Web and Research does the same inside a chat, one message at a time.
- Approvals that say what will happen. Each approval opens with one plain sentence, such as “Run ‘npm test’ in ferry” or “Write report.md (2.1 KB) in Documents”, with the tool, arguments and folder still shown below.
- An Inbox, and runs that wait. Everything your workflows are waiting on sits in one Inbox: questions, reviews, outside actions and command approvals. The rail shows a dot, and new requests raise a system notification. A run waiting for you carries on after you close and reopen Conduit, with its original deadline, and the time spent waiting does not count against its time limit.
- Tray is opt-in. To keep schedules running with the window closed, Conduit can stay in the tray and start there when you sign in; both are off by default.
Home
The app opens on Home. See the Home docs.
- One ask box. Ask for a deck and Slides opens with a storyline; anything else starts a chat. Quick starts sit underneath.
- Needs you. Workflow approvals, workflow questions, and memory suggestions waiting on you in one list, hidden when nothing is. Workflow requests open the Inbox.
- Fresh from your workflows. The results of recent runs that you haven't opened yet.
- Pick up where you left off. Your latest chats, decks, drafts, and apps.
- A tile per area, with a live count. An empty area shows an example to try instead.
- Go anywhere from the keyboard. Ctrl+1 to Ctrl+9 (⌘ on macOS) open Home, Chats, Slides, Writing, Apps, Workflows, Inbox, Agents, and Context. Manage has no number; the rail, Home, and the command palette's "Go to" entries reach it.
Models & providers
Conduit talks to a model provider directly with your own key — there is no Conduit account and no Conduit-hosted model. See adding a provider for setup.
- Twenty-three built-in providers. Anthropic, OpenAI, Google Gemini, xAI, Z.ai, Moonshot AI, Qwen, MiniMax, DeepSeek, Mistral, Groq, Together AI, Fireworks AI, Cerebras, SambaNova, DeepInfra, Novita AI, NVIDIA NIM, OpenRouter, OpenCode Zen, Ollama, LM Studio, and any OpenAI-compatible endpoint.
- Custom base URL on seventeen of them. Anthropic, OpenAI, Ollama, LM Studio, OpenAI-compatible, xAI, Z.ai, Moonshot, Qwen, MiniMax, Together AI, Fireworks AI, Cerebras, SambaNova, DeepInfra, Novita AI, and NVIDIA NIM take an editable endpoint, so a LiteLLM proxy or compatible API works too. Gemini, OpenRouter, OpenCode Zen, Groq, DeepSeek, and Mistral use a fixed endpoint.
- Model families beyond the dedicated adapters. Grok, GLM, Kimi, Qwen, and MiniMax each have a first-class provider (xAI, Z.ai, Moonshot, Qwen, MiniMax). Anything else comes through OpenRouter, OpenCode Zen, an OpenAI-compatible endpoint, or Ollama and LM Studio.
- Local models with no key. Ollama and LM Studio are supported directly, plus any OpenAI-compatible server you run yourself, such as vLLM or SGLang.
- Bring your own key. Credentials live in the OS keychain and requests go straight from your machine to the provider — Conduit never proxies a request or resells a token.
- Local-only mode. One toggle refuses every cloud provider — only a local model such as Ollama or LM Studio can answer — and turns off web search, cloud document indexing, and page network access. It's on by default for a fresh install; choosing a cloud provider during setup switches it off.
- Switch mid-conversation. ⌘⇧P cycles the active provider without starting a new chat.
- Vision. Image attachments go to models that accept them, DeepSeek V4 Flash among them. A text-only model gets the image dropped with a note rather than a faked answer.
- Documents in a message. Attach a PDF, Word, Excel, PowerPoint (
.pptx), text, Markdown, CSV or HTML file and every model gets its text, extracted on your machine; models that read PDFs (Claude, OpenAI, Gemini, and OpenRouter models that list file input) get the PDF too. The content goes to the chat provider you picked. See Documents. - Model search. The model menu opens with a search field across every provider's list; Enter takes the first match, and each provider's models are listed alphabetically.
- Reasoning effort. Auto, Low, Medium or High, per chat or as a default; lower is faster and cheaper on thinking models, and Auto leaves the model's default. Sent only to models that accept it, and mapped for OpenAI, OpenRouter, Anthropic and Gemini — not DeepSeek.
- Claude prompt caching. Requests to Anthropic, and to Claude models on OpenRouter and OpenCode Zen, mark the tools, system prompt and conversation so far for Anthropic's prompt cache. Cache reads and writes are priced at their own rates in Usage & Cost, so what it saves shows there.
- Extended reasoning display. Anthropic's thinking output renders in the conversation.
Documents
Collections of your own files, searchable from the chat. See the documents page for a screenshot of how sourced passages appear.
- Eight file types, parsed on-device. Plain text, Markdown, CSV, HTML, Word (
.docx), Excel (.xlsx), PowerPoint (.pptx), and PDF. A PowerPoint file is read slide by slide, with tables, chart data and speaker notes; macro-enabled.pptmfiles are refused. For HTML, the page's title and text are read and scripts and styles are left out. There is no OCR step — a scanned PDF with no text layer yields nothing. - Or attach one file to a message. + → Attach files…, paste, or drop it on a chat — in Chats a drop anywhere attaches, elsewhere it adds to Documents. No collection or embedding needed: every model gets the text, extracted on your machine, and models that read PDFs get the PDF too, so a scanned page can come through. Each attachment says how it will be sent, or why it won't be. The content goes to the chat provider you picked. A PDF over 20 MB goes as text only; a document's text is cut at 150,000 characters.
- Hybrid search. Vector similarity plus SQLite full-text search, merged by reciprocal rank fusion, both running against the same local database as your chats.
- Sourced answers. The files an answer drew on appear as chips on your message; click one to read the exact passage that was used.
- Four embedding providers. OpenAI, Gemini, OpenRouter, or Ollama — Anthropic cannot embed. A document's full text is sent once to whichever provider you pick; consent is asked per provider, remembered, and can be withdrawn.
- Local-only mode forces Ollama. With it on, cloud embedding is refused outright, so a collection stays entirely on your machine. Retrieved passages still go to whichever chat provider answers.
Images & artifacts
- Image generation on three providers. OpenAI, Gemini, and OpenRouter only. The finished image is saved locally as an artifact rather than left as a link that expires. Images are billed by the provider, and Conduit asks once before the first one.
- Artifacts render beside the chat. Text, code, JSON, Markdown, Mermaid diagrams, KaTeX maths, generated images, and sandboxed HTML all open in a side panel, kept out of the message scroll.
- Interactive HTML is sandboxed, not just warned about. A frame with scripts allowed and everything else forbidden — no same-origin access, a content policy that blocks the page's own network requests,
srcdocdelivery so it has no origin to trade on, and no bridge back into the app. Forms and buttons work. - Pages can fetch live data, from sites you allow. A page that wants a currency rate or a weather feed asks first: a banner names the site, and you allow it once, always for that page, or not at all — or let that page reach any public site. Conduit makes the request itself, over https only, with no cookies or credentials and no access to local or private addresses, and every request is logged. Permissions belong to one page, not the app. Refused entirely in local-only mode, and one switch in Settings → Privacy & data turns it off.
- Export to disk, with an optional metadata sidecar. There is no version history — see Not built.
Agents & tools
Run control, transparency, and consent for what the model does on its own.
- Run control. Queue a follow-up while a turn is still running, or interrupt it mid-loop and redirect it — the run doesn't have to finish before you can act.
ask_userforms. When the model needs a detail from you, it can put a real form in front of you — text fields, dropdowns — instead of guessing.- Live progress. A segment timeline shows thinking, each tool call, and the answer in the order they actually happened across rounds, with elapsed time and stall detection. A document being written shows its line count, size, and a "still working" note.
- An inspector for what happened. Once a reply is done, its steps fold into one line — "2 steps · 1 site" — that opens the inspector beside the chat. Its Activity tab lists every tool call, search, and site contacted, with timing and status; Sources lists the search results and document passages the answer used. Anything that needs you — an approval or a question — still appears in the chat.
- Chats that say when they need you. A chat you've left shows "Running" or "Needs you" in the chat list, so a turn waiting on an approval doesn't sit unnoticed.
- Remembered tool approvals. Approve a tool for this chat or for good. Sensitive tools are the exception — never remembered, always asked.
- Skills (
SKILL.md), enabled per conversation. Metadata is read first and the full instructions load only when you switch a skill on. Conduit does not execute a skill'sscripts/— they are listed as documentation, never run. - Make a skill from a chat, or import one from GitHub. The command palette's “Make a skill from this chat” has the current model draft a name, description and instructions from what you asked; you edit the draft and nothing is saved until you choose Save. Tool output and document text are never sent to the draft. Settings → Skills also takes a
github.com/owner/repolink or a link to a folder in it: only that folder is unpacked, and scripts in a skill are never run. - Your agents, defined once. On the Agents page you make an agent once — name, instructions, model, tools, the agents it may hand work to, and a token limit — and use it in chat, in any workflow, or on a schedule. Editing it changes it everywhere. Researcher, Writer, Reviewer and Data analyst come with a new profile. An agent that Remembers keeps its own notes between runs, which you can read and edit, and one that Reads can search the knowledge collections you pick.
- Agents in chat, side by side. Type
@to pick one of your agents for a turn, and choose in Settings → Agent which agents the assistant may hand work to. Each hand-off runs on that agent's own model and shows as one line with its status, tokens, cost and time; several run at the same time (3 by default, 1 to 4 in settings), except helpers that change things, which go one at a time. Open one to watch its task, tool calls and cost, then Stop it or Redirect it while it works. - Routines and reviews. “Run on a schedule…” on an agent's card gives it a task to do daily, on weekdays, every few hours, or when a feed or folder has something new, with each answer in Results. The document panel can send an HTML, Markdown or text document in a chat to your Reviewer agent, and the chat revises the document if it finds problems.
- User-approved memory. The model can propose a fact through a tool, but it lands as pending — reaching no future prompt until you save it. Everything kept is a plain list on the Memory page you can read, edit, or delete. Memory is on by default, and stored items are encrypted only once encryption at rest is switched on.
- Context gauge and automatic compaction. The gauge estimates how full the next request will be, not summed turn totals. Compaction is on by default at 90% full: older turns are journaled into a summary while the most recent ones stay verbatim, and raw messages are never deleted.
- Built-in tools, no setup. Web search and fetch, document write and edit, clipboard, calculator, time, UUID, and random values — available without adding a connector. The page fetch reads public https sites only, never your own network, and reads PDFs up to 20 MB.
- Web search, off by default. Hosted search runs on the official Anthropic, OpenAI, or Gemini endpoint, or through OpenRouter's built-in search — a custom base URL strips it. Every other model searches through a local backend: Exa (the default; free and rate-limited without a key, and queries go to Exa under its terms), Tavily, Brave, your own SearXNG, or DuckDuckGo, with a link in Settings → Web search to get a key. It's disabled entirely in local-only mode.
- Workspace folder tools, opt-in. Off by default and gated behind an explicit acknowledgement, they give the model read, write, edit, glob, and grep inside one folder you choose — path-sandboxed, with no shell execution. Reading covers the text of PDF and Word files in that folder too.
Connectors
MCP — the Model Context Protocol — for bringing your own tools in on your own terms.
- Local stdio servers. Supervised child processes with restart, backoff, concurrency limits, and per-call timeouts.
- Remote MCP over streamable HTTP. The current spec, with a fallback to older ones. The deprecated SSE-only shape is deliberately rejected rather than silently accepted.
- OAuth, with tokens in the keychain. Sign in to a remote server that uses OAuth; Conduit identifies itself with a published client metadata document, and tokens are stored in the OS keychain, never held by the interface.
- Official registry search. Search
registry.modelcontextprotocol.ioand install a remote server in one click. - Prompts and resources in the composer. A prompt picker fills in the arguments a server prompt declares and puts the result in the composer to edit before you send it. A resource picker attaches a resource to the next message only. A resource that tries to issue instructions is refused and named, not silently followed.
- A tool asks before it runs unless its server marks it read-only. Read-only tools run without interrupting you; every other tool stops and asks first, and you can remember an approval for one chat or always. Tool output is size-capped and redacted, with a structural guard against it being replayed as an instruction.
- “Only reads” for tools that don't say. In Context → Connectors, each tool shows what it declares. A tool that declares nothing asks before every call; tick Only reads for one you trust to only look things up and it runs without asking in chats, agents and workflows. A tool that says it changes things, or is sensitive, keeps asking.
- Many connectors, without the clutter. When a chat would offer more than 20 connector tools, or their descriptions run long, the model gets a tool-search tool instead of the whole list and loads only the tools it finds. A connector you name in your message is offered directly.
Conversations
- A grouped side rail. Home; Chats, Slides and Writing; Apps and Manage; Workflows, Inbox and Agents; Context, which holds Documents, Library, Connectors and Memory as tabs; and Settings at the bottom. Each opens as a full page, and a reply keeps running while you're on another one. The place you are in shows a solid icon, and Context reopens on the tab you used last.
- Your apps as cards. The Apps page shows each app with its name, category, when you last opened it, what it does, and whether it uses the internet or your model, with a search box and a New app button. Manage keeps the details, network access and starter apps.
- Examples to look at. A new profile starts with a live weather app, the chat that built it, a plan for a rainy Saturday, an 8-slide deck on how LLMs work and a Morning briefing with a result waiting, in your language and with no API key. Each is marked Example, and Settings → Privacy & data removes them all.
- Easier to read. No text in the app is smaller than 12px, secondary text is brighter in dark, and light has stronger lines and text on white cards.
- Ideas to start from. Twenty-one hand-written, tested starting points — a pomodoro timer, a budget tracker, flashcards, a game of Snake — in seven groups, under the message box in a new chat. Picking one fills in a prompt you can edit before sending. Ideas that need something you haven't set up say so. Which ones you've tried is remembered on your device only.
- One + menu in the composer. Attach files, web search, a workspace folder, documents, skills, and connector prompts and resources from one button; whatever is switched on shows as a chip you can remove.
- Full-text search across every conversation, surfaced in the ⌘K command palette alongside model and artifact switching.
- Edit any message and resend. Editing the latest message rewrites it in place. Editing an earlier one always forks a new conversation that branches before that message, and the original is kept.
- Retry keeps every answer. Each retry of the last reply is kept as a version, and a
‹ 2 / 3 ›switcher under the reply moves between them; the chat continues from the one showing. An earlier reply shows its versions but cannot switch, because the messages after it were written from the one it has. - Retry with another model. The arrow next to Retry opens the model menu, and the model you pick answers the last question again, for that reply only: the chat keeps its own model. The command palette has “Retry with another model…” too.
- Compare answers side by side. Compare opens the versions in the panel on the right, each with the model that wrote it, its answer and its cost. “Use this one” picks a version for the last reply, and “Fork from this version” starts a new chat that continues from any version. A reply records the model that wrote it, so its model line stays right after the chat is reopened.
- Continue a cut-off reply. When a reply stops at the output limit, a Continue button under it picks up where the model stopped, in the same message, with no new message from you. It works with every provider.
- Projects: the same instructions, documents and skills in every chat. Set them once in a project's settings, from its ⋯ menu in the chat list. Instructions stack, each labelled: Settings, then the project, then the chat. Collections and skills are the project's plus the chat's own, and the chat settings popover shows what comes from where. A project is read when you send a message, so editing it changes every chat in it, and a chat moved in or out follows from its next message. See Projects.
- A working folder per project. New chats in a project start in its folder, so the model can read and write files there with the opt-in workspace tools: path-sandboxed, with no shell. The folder's
AGENTS.md, if it has one, goes with every message. Chats already in the project keep the folder they started with. - Pin, archive, and projects one level deep. Every conversation has a menu for renaming, pinning, archiving and moving it; projects have no nesting, and a project with no settings is just a folder.
- Export a conversation as Markdown or JSON.
- Saved prompt library. Folders, tags, and
{{variable}}templating for prompts you reuse. - Usage and cost tracking — estimates. Per-day cost and token charts computed locally from your own usage, counting every model call — chats, Research runs, workflow steps. Prices come from your own overrides first, then OpenRouter's listed prices for its models, then a snapshot of the models.dev catalog bundled with each release (Conduit never contacts models.dev). A model with no known price says No price instead of $0, and the total says how many it leaves out; local models cost nothing. Actual billing may differ.
- Daily spend alert, off by default. Set an amount in Settings → Usage & Cost and, after a reply that takes today's estimated spend past it, a notice shows the estimate — at most once a day. It never stops anything, and models without a price aren't counted. Days are UTC, like the usage chart.
Privacy & security
The full statement, including every network flow and its default, is on the privacy page. The highlights:
- The trust boundary. The interface sends a typed request and receives typed events — it never holds a key or opens a socket. Keys, the database, and the HTTP client live only in the Rust core.
- Keys in the OS keychain. Windows Credential Manager, macOS Keychain, or Linux Secret Service, referenced by name and never held in Settings as a value.
- Local SQLite storage. Conversations, attachments, documents, and artifacts live in a database and content-addressed blob store on your machine — no Conduit server to upload them to.
- Local-only mode. One toggle refuses cloud providers on every path, and switches off web search, cloud document indexing, and page network access. Use it with Ollama or LM Studio and switch off update checks, and the app makes no internet request at all.
- Optional encryption at rest — off by default. AES-256, keyed from the OS keychain, covering attachment blobs, artifact content, document chunks, memory items, saved prompts, per-chat custom instructions, and tool results. It does not cover message text, the event log, compaction summaries, or the search index. If the key goes missing while encrypted data exists, the app refuses to start rather than fall back to plaintext.
- No telemetry. No background analytics or usage reporting. Diagnostics exports are manual, redacted, and local files — nothing is sent anywhere on its own.
- Signed update payloads. Every update is Ed25519-signed and verified before it's applied. The installers themselves are not OS code-signed, so Windows SmartScreen and macOS Gatekeeper still warn on first launch — see the changelog.
- Optional automatic updates. The default policy is manual — nothing is checked until you press Check now. Automatic checking and installing are opt-in; an automatic install is staged and applied on quit, never by restarting the app on its own, and isn't offered for the Linux
.debbuild.
Personalisation
- One design, dark and light. Choose dark, light, or following your system. A labeled rail gives every feature one place.
- Your main colour. Pick the accent for each mode in Settings → Appearance; a colour that would be hard to read is refused.
- Interface in eight languages. English, German, Spanish, French, Japanese, Korean, Brazilian Portuguese, and Simplified Chinese, with dates, numbers, and file sizes to match. Changing it doesn't change the language the assistant replies in.
- Settings, in sections. Usage & cost and Updates have their own sections. About shows the version, links to the website, source code and issue tracker, the license, and where the app keeps its files on this machine. The command palette opens each section directly.
- Small windows. Below 720px wide the rail becomes a bottom tab bar (Home, Chats, Apps, Inbox, More), two-column pages stack, and on touch screens hover-only actions stay visible with larger tap targets. The window can be as narrow as 360px, and the rail turns compact in short windows.
- Text size. Compact, default, or comfortable, applied across the interface.
Voice
- Read aloud. A speaker button under each reply reads it out with your system's voices, in the reply's language. Click it again to stop.
- Dictation. A microphone beside Send records you; click it again, or wait for the five-minute limit, and the text lands in the composer. Escape discards the recording. Your voice goes to the transcription provider you choose in Settings → Voice: OpenAI, OpenRouter or Groq, with your own key. Dictation is off in local-only mode.
Keyboard
- Command palette (⌘K), modal. Type
>for commands,@for artifacts,/for models, or plain text to search every message you've sent. - Shortcuts sheet opens with Ctrl+/ on Windows and Linux, ⌘/ on macOS, listing the full set: ⌘N new chat, ⌘, settings, ⌘\ sidebar, ⌘J document panel, ⌘⇧E widen the artifact panel, ⌘⇧P cycle provider, ⌘⇧W web search, ⌘⇧F fork, ⌘⇧C copy the last reply, Esc to close a panel or stop a response.
- Quick ask shortcut, off by default. Set one in Settings → Chat and it opens a small ask box from anywhere; Enter sends the question to the app, which opens it like the Home ask box. Esc or clicking away hides it. Not available on Linux Wayland, where global shortcuts generally don't work.
- Settings search. Every setting is findable by name, not just by which section it's filed under.
Platforms
- Windows 10 and 11 — x86_64 installer.
- macOS 11 (Big Sur) and later — separate builds for Apple silicon and Intel.
- Linux — x86_64, as a
.debpackage or an AppImage. - No feature differences between platforms. Every capability above works the same on all three, except that automatic update installs aren't offered for the Linux
.deb.
White-label
Full status table and details on the white-label page.
- Runtime branding ships today. From Settings → Branding, a stock install can change its product name, tagline, in-app logo, window title, and around eighteen accent and surface colours — reversibly, in minutes.
- Build-time branding is early access. Producing your own installer — with your own app icon, installer name, bundle identifier, bundled fonts, and update channel — is built from source today and open to design partners; there is no pre-built branded installer to download yet.
- Not yet built: shipping a build pre-wired to your own connectors, and OS-signed and notarized branded builds — both are in development, not shipping.
Not built
Said plainly, so nothing above is mistaken for a promise about these:
- No cloud sync or multi-device access — every install is its own, independent database.
- No account, login, or hosted service of any kind.
- No artifact version history — it was built, then deliberately removed. An artifact is a single payload by design.
- No Azure OpenAI or AWS Bedrock provider.
- Installers are not OS code-signed or notarized, so Windows and macOS both warn on first launch.
- The assistant does not switch its reply language automatically — picking an interface language changes the UI, not the language the model replies in.
Every claim above is checked against the release it describes. See what changed release by release, what stays on your machine, or the full documentation.