Show an agent a UI problem
Capture one reproduction, assemble the evidence, and give your coding agent a concrete investigation.
Eavesy keeps the things you'd otherwise re-explain (clipboard text and images, screenshots, dictation, recordings) as files and reusable text on your Mac. Copy it into Claude, Cursor or ChatGPT, or let an agent read synced recording text over hosted MCP.
Hold ⌘⇧D and talk. Set Talk to fn in Settings if you prefer.
Capture from the tools you use. Send context to the tools you choose.
A native Mac app that keeps what you save as plain Markdown, so your agent gets the full story instead of a vague prompt.
Paste text or an image from your clipboard, take a screenshot, dictate a thought, or record your screen and voice. Eavesy saves each one as a capture on your Mac, so the context is ready when you are.
Signed-in recordings receive a cloud transcript and Markdown summary. Dictation produces text to paste; its signed-in cloud copy can be cleaned up. Copy it into Claude, Cursor, ChatGPT, or any agent instead of re-typing the context from memory, or let an MCP-capable agent fetch it itself.
The Mac app writes each capture to a normal folder as Markdown, JSON, and media. Point an agent at the folder, open it in Obsidian, or grep it. Dictation and recordings are sent to the cloud for transcription once you sign in, and screenshots and images sync to your other signed-in Macs; clipboard notes stay on your Mac. You can bring your own OpenRouter account for the AI work.
What you can drop in
Eavesy holds the text, images, and recordings your prompt would otherwise lose, as Markdown you can copy in a second.
Open the launcher with your global hotkey and press ⌘V to save the text or image on your clipboard as a capture you can find and copy again later.
Grab the screen and keep it next to the rest of your context for a bug report or a coding agent.
Speak into the app you are using. On-device dictation works without an account when recognition is available; pasting needs Accessibility permission.
Record your screen, system audio, and mic. Sign in to get a transcript and a Markdown summary when you stop.
Capture starts only when you start it, files stay on your disk, and you decide what gets shared.
Read local captures as files, or connect the hosted MCP server to search synced recordings and dictations. Local screenshots and notes are not uploaded to MCP.
Open the launcher with a hotkey, then paste your clipboard, take a screenshot, dictate, or start a recording.
Audio becomes a transcript and a structured Markdown summary. Text and images are saved as they are.
Copy the Markdown into any AI agent, point it at the folder, or let it read synced recordings over hosted MCP.
Put your context to work
See what to record, how to check the source, and what to hand your agent. Use the guides and editable Markdown templates with your own evidence.
Capture one reproduction, assemble the evidence, and give your coding agent a concrete investigation.
Check both audio sources, preserve the decisions, and carry confirmed tasks into your project notes.
Find the recording, check the supporting passage, and retain what is still unknown.
A local, AI-first assistant for your Mac. You drop things into it (clipboard text and images, screenshots, dictation, and recordings) and it saves them as local files and readable text that is quick to paste into Claude, Cursor, ChatGPT, or any agent, with hosted MCP for synced recordings and dictations.
Copy the Markdown for a capture and paste it into any agent. Signed-in recordings can include a transcript and summary; dictation supplies text to paste. You can also point an agent at the capture folder on your disk, or connect the read-only MCP server so a compatible agent can search your library itself.
You can paste any text or image from your clipboard into the Mac app and it is saved as a note or image capture. A dedicated flow for forwarding messages and orders from other apps is not available yet.
Yes. The Mac app records your screen, system audio, and microphone with no bot joining the call, and the web app has a browser recorder. Recording is one of the inputs, not the whole product. On-device live captions are available with Speech Recognition permission. Signed-in recordings also receive cloud transcripts and summaries after processing. You remain responsible for following consent rules.
Pasting clipboard text or images and taking screenshots on your Mac works without one. On-device dictation, captions, and local recording work without an account when the required permissions and speech recognition are available. Sign in for cloud transcription, summaries, and hosted search and MCP access to synced recordings.
With the Mac app, each capture is a folder on your disk containing Markdown, JSON, and media files. Dictation and recordings are also uploaded to a private cloud copy for transcription, search, and recovery, and screenshots and images sync between your signed-in Macs. Clipboard notes are not uploaded, and signed out, nothing leaves your Mac. You can connect your own OpenRouter account to run the AI on your own credits.
The Mac app supports Apple Silicon Macs on macOS 13 or later (screenshots need macOS 14). The web app and browser recorder work in a modern browser. Other platforms are not available yet.
Drop in what matters and get Markdown you can hand to any AI. Free for Mac.
Free · macOS 13+ on Apple Silicon · No credit card