# Talkthrough

A local MCP server that turns a screen recording or a public video link into a searchable transcript, key frames, OCR text and wall-clock timestamps.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 32
- Author: korovin-aa97
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/korovin-aa97/talkthrough-mcp
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/mcp/talkthrough

## What it does

Point it at a `.mov` or `.mp4` with narration, or at a public URL. It indexes the file into a transcript, saved frames, text read off the picture and marks tied to real playback time. Processing happens on your own machine; nothing gets uploaded.

The server ships 9 tools and 6 prompts, with speaker diarization as an option. The plugin adds slash commands such as `/talkthrough:bug` and `/talkthrough:triage-recording`, plus a triage subagent and a skill that walks the model through the flow. The agent drafts an issue but never files it; posting stays with your own `gh` or tracker.

You need `uv`. `uvx` resolves the Python version and the server environment for you. Expect the first `process_media` call to take a while: speech and OCR models are fetched then. No system ffmpeg? A bundled build is used instead. Updating the plugin can rebuild the environment and pull that build again. Register the server once only, either as a bare server or as the full plugin, not both.

## Requirements
- Needs: Python and uv (commands run via uvx)
- Environment variables: `TALKTHROUGH_WHISPER_MODEL`, `TALKTHROUGH_OCR`, `TALKTHROUGH_OCR_LANG`, `TALKTHROUGH_HOME`, `TALKTHROUGH_MAX_DOWNLOAD_BYTES` — keys and tokens: ask the human for the values, never commit them or put them in shared files.

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add --env TALKTHROUGH_WHISPER_MODEL=<TALKTHROUGH_WHISPER_MODEL> --env TALKTHROUGH_OCR=<TALKTHROUGH_OCR> --env TALKTHROUGH_OCR_LANG=<TALKTHROUGH_OCR_LANG> --env TALKTHROUGH_HOME=<TALKTHROUGH_HOME> --env TALKTHROUGH_MAX_DOWNLOAD_BYTES=<TALKTHROUGH_MAX_DOWNLOAD_BYTES> --transport stdio talkthrough -- uvx talkthrough-mcp
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from talkthrough-mcp <command>

### Codex

**Platform CLI**
```bash
codex mcp add talkthrough --env TALKTHROUGH_WHISPER_MODEL=<TALKTHROUGH_WHISPER_MODEL> --env TALKTHROUGH_OCR=<TALKTHROUGH_OCR> --env TALKTHROUGH_OCR_LANG=<TALKTHROUGH_OCR_LANG> --env TALKTHROUGH_HOME=<TALKTHROUGH_HOME> --env TALKTHROUGH_MAX_DOWNLOAD_BYTES=<TALKTHROUGH_MAX_DOWNLOAD_BYTES> -- uvx talkthrough-mcp
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from talkthrough-mcp <command>

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "talkthrough": {
            "command": "uvx",
            "args": [
                "talkthrough-mcp"
            ],
            "env": {
                "TALKTHROUGH_WHISPER_MODEL": "<TALKTHROUGH_WHISPER_MODEL>",
                "TALKTHROUGH_OCR": "<TALKTHROUGH_OCR>",
                "TALKTHROUGH_OCR_LANG": "<TALKTHROUGH_OCR_LANG>",
                "TALKTHROUGH_HOME": "<TALKTHROUGH_HOME>",
                "TALKTHROUGH_MAX_DOWNLOAD_BYTES": "<TALKTHROUGH_MAX_DOWNLOAD_BYTES>"
            }
        }
    }
}
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from talkthrough-mcp <command>

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "talkthrough": {
            "type": "stdio",
            "command": "uvx",
            "args": [
                "talkthrough-mcp"
            ],
            "env": {
                "TALKTHROUGH_WHISPER_MODEL": "<TALKTHROUGH_WHISPER_MODEL>",
                "TALKTHROUGH_OCR": "<TALKTHROUGH_OCR>",
                "TALKTHROUGH_OCR_LANG": "<TALKTHROUGH_OCR_LANG>",
                "TALKTHROUGH_HOME": "<TALKTHROUGH_HOME>",
                "TALKTHROUGH_MAX_DOWNLOAD_BYTES": "<TALKTHROUGH_MAX_DOWNLOAD_BYTES>"
            }
        }
    }
}
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from talkthrough-mcp <command>

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add -e TALKTHROUGH_WHISPER_MODEL=<TALKTHROUGH_WHISPER_MODEL> -e TALKTHROUGH_OCR=<TALKTHROUGH_OCR> -e TALKTHROUGH_OCR_LANG=<TALKTHROUGH_OCR_LANG> -e TALKTHROUGH_HOME=<TALKTHROUGH_HOME> -e TALKTHROUGH_MAX_DOWNLOAD_BYTES=<TALKTHROUGH_MAX_DOWNLOAD_BYTES> talkthrough uvx talkthrough-mcp
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from talkthrough-mcp <command>

## When to use it
- turn a recorded bug into a written transcript
- grab the frame where an error shows up
- read text off screen and map it to a timestamp
- draft an issue from a meeting recording
- index a public demo link without downloading it by hand

## Example prompts
- Process `checkout-bug.mp4` and draft an issue with frames and timestamps
- Run this video through talkthrough and find where the error appears
- Transcribe the recording and pull the on-screen text

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [browser-use](https://affpapa.org/en/nevibecode/mcp/browser-use.md) — MCP server · ★ 119160
- [Codebase Memory](https://affpapa.org/en/nevibecode/mcp/codebase-memory.md) — MCP server · ★ 47392
- [mcp-server](https://affpapa.org/en/nevibecode/mcp/mcp-server.md) — MCP server · ★ 43872

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md