get_transcripts_relativeLABSCO SUMMARY
Your own recall, queryable.
Pairing audio with screen OCR is what makes it more than a transcript search: what was said in a meeting and what was on screen during it are two halves of the same memory, and both are addressable by the same time window.
What it is
A reader for the Rewind.ai SQLite database on your Mac. Rewind captures audio transcripts and screen OCR continuously; this exposes both to an assistant, so a question about a meeting last Tuesday or what was on screen during it becomes a query rather than a scroll through a timeline.
What you get
- Transcripts —
get_transcripts_relativefor a period such as 1hour or 1week,get_transcripts_absolutefor a specific window such as a 3 PM meeting, andget_transcript_by_idto pull one session's full text after the first call showed a preview search_transcripts— keyword and phrase matches with context snippets across sessions, for finding where a topic came up rather than summarising a known meeting- Screen OCR —
get_screen_ocr_relativeandget_screen_ocr_absolutefor everything visible in a period, organised by application and timestamp, andsearch_screen_ocrfor keywords that appeared on screen - Application discovery —
get_ocr_applications_relativeandget_ocr_applications_absoluteshow which applications were active and how busy, so you can filter OCR content before pulling it get_activity_stats— audio recordings, screen captures and application usage over a period- Voice filtering on the CLI side:
--speech-sourceseparates your own voice from other speakers, with text, JSON and audio export
Requirements
A Mac running Rewind.ai, since it reads that application's own database. Python 3.6+, then pip install . from a clone. Connection details go in a .env — DB_PATH pointing at the Rewind database and DB_PASSWORD — kept in your project directory or at ~/.rewinddb.env, with --env-file for anywhere else. The MCP server is mcp_stdio.py, which takes the same --env-file argument. Everything it can reach is your own recorded activity, so treat the database path and password like any other credential.
Setup effort
Build from source — git clone https://github.com/pedramamini/RewindMCP.git
USE CASES
01
Summarising a meeting you half remember
Pull the transcripts for that exact window, then fetch the full text of the session when the preview is not enough.
02
Finding where something was mentioned
Search the transcripts for the name or topic and get context snippets across sessions rather than one meeting.
03
Recovering what was on screen
Discover which applications were active in the period, then pull the OCR content from the one you meant.
04
Tracking down a value you saw but did not save
Search the screen OCR for the text and find the moment it was visible.
05
Reviewing how a period was spent
Read the activity statistics across audio, screen captures and application usage for the window.
06
Collecting your own voice as clean text
Export by day with the speech source set to your own voice, so other speakers are excluded from the text.
10 tools
Listed in project docs. Your client may show a different list.
get_transcripts_absolutesearch_transcriptssearch_screen_ocrget_screen_ocr_relativeget_screen_ocr_absoluteget_ocr_applications_relativeget_ocr_applications_absoluteget_activity_statsget_transcript_by_idBEFORE YOU INSTALL
What to know first
Tools you get
10 tools named in the project's documentation.
Requires
- Python 3.6+
- DB_PASSWORD in your environment
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
You clone this one yourself.
Server setup
TERMINAL
git clone https://github.com/pedramamini/RewindMCP.git
cd RewindMCP
pip install .MCP Clients
Any MCP client
ANY MCP CLIENT · JSON
{
"mcpServers": {
"rewinddb": {
"command": "python",
"args": [
"<YOUR_PATH_TO_REWIND_MCP>/mcp_stdio.py"
],
"env": {
"DB_PATH": "<YOUR_DB_PATH>",
"DB_PASSWORD": "<YOUR_DB_PASSWORD>"
}
}
}
}
