Read your saved Instagram posts and have Reels transcribed and summarised locally, from a personal account.
You saved a Reel in Japanese three weeks ago and cannot remember what it said. Instagram Obscura is built for exactly that: point your assistant at a Reel or your saved grid and get the original-language transcript, timestamps, sampled frames and an explanation in your own language.
The Instagram Obscura MCP server works through a browser session on a personal account. You log in yourself, in a visible Chrome window opened by Playwright. Browsing after that happens inside Obscura, a separate browser engine. No Business or Creator conversion and no Graph API app are involved.
There are 22 tools in total. Processing uses FFmpeg and runs as bounded jobs with a verified cache, all on your machine.
Expect a hands-on install. You need Python 3.12 or newer, Google Chrome, FFmpeg with ffprobe, an Obscura binary and, for speech, a local Whisper model. It installs from source into a virtual environment with python -m pip install ".[analysis]", and the instagram-mcp-doctor command checks what is missing. It is not on PyPI.
Two environment variables tell it where things live: INSTAGRAM_OBSCURA_BINARY and INSTAGRAM_WHISPER_MODEL. Use absolute paths throughout, including for the instagram-mcp executable itself.
Ask the assistant to run instagram_health, then instagram_login_start. Complete login and two-factor authentication yourself in the window that opens, then have it call instagram_login_finish and instagram_account_info to confirm the account. Never paste a password or 2FA code into the chat.
This is unofficial and experimental, and it is not affiliated with Meta. Features can break whenever Instagram changes its pages. Windows is the only platform with real-engine validation. Messaging, comments and publishing drive the visible UI rather than an API, and real sends were not part of release testing, so ask for specific actions and check the result yourself.
Transcription quality varies by language and recording; there is no OCR, and sampled frames can miss quick moments. File uploads are off by default, and enabling them also widens Obscura's local file access.
For other platforms, see Reddit Content or LinkedIn Browser, or browse the social media servers we list.
Check the linked publisher documentation and your connected client's tool list. Names and availability can vary with the server version and account permissions.
| Tool | What it does |
|---|---|
| instagram_health | Check that the server and its dependencies are ready. |
| instagram_login_start | Open a visible browser window so you can log in and complete 2FA yourself. |
| instagram_login_finish | Complete the login handoff after you have signed in manually. |
| instagram_account_info | Show which Instagram account is active in Obscura. |
{
"mcpServers": {
"instagram": {
"command": "/absolute/path/to/.venv/bin/instagram-mcp",
"args": [],
"env": {
"INSTAGRAM_OBSCURA_BINARY": "/absolute/path/to/obscura",
"INSTAGRAM_WHISPER_MODEL": "/absolute/path/to/whisper-small"
}
}
}
}Use absolute paths from your own virtual environment. On Windows the command is C:/absolute/path/.venv/Scripts/instagram-mcp.exe. Requires Python 3.12+, Chrome, FFmpeg and an Obscura binary.
| Variable | Description | Required |
|---|---|---|
| INSTAGRAM_OBSCURA_BINARY | Absolute path to the Obscura browser engine binary used for browsing after login. | Yes |
| INSTAGRAM_WHISPER_MODEL | Absolute path to the local multilingual Whisper model used for Reel speech transcription. | Optional |
Search built for AI, not humans — semantic web search that returns model-ready content, plus code context.
Put 6,000+ pre-built scrapers at your assistant's fingertips through one MCP endpoint.
XActions - The Complete X/Twitter Automation Toolkit. Scrapers, MCP server for AI agents, CLI, and browser scripts. No API required. Open source by
MCP server to gather usernames from LinkedIn using Serper
Turn your Android phone into an MCP (Model Context Protocol) server. AI agents and desktop scripts can call your phone for live data and actions over
OmniX is the fast, affordable X (Twitter) API — one key for full read & write access (tweets, users, DMs, media, articles), no official X developer