dothething
an autonomous AI agent: you describe the thing, it does the thing.
Dothething is an autonomous local AI agent that accepts task descriptions in plain English and executes them until completion, but also functions as an MCP server via its --browsermcp mode, exposing browser automation and search tools to external AI clients. The agent handles multi-step workflows including web research through SearXNG, browser automation with anti-fingerprinting via Camoufox, file operations, shell commands, and email integration. When run as an MCP server, it provides dtt_browser, dtt_fetch, dtt_browser_agent, and dtt_browser_session tools that other agents can invoke for web scraping, captcha solving, and maintaining persistent browser sessions with saved logins.
Key Features
Use Cases
- 01Providing Claude Desktop or other MCP clients with advanced browser automation and anti-bot capabilities
- 02Scraping content from login-protected sites by reusing saved browser sessions across multiple agent runs
- 03Automating research tasks that require bypassing Google's bot walls and captcha challenges
- 04Building multi-agent workflows where one agent delegates web scraping to dothething's browser tools
- 05Creating persistent browser profiles for repeated interactions with the same authenticated services
- 06Extending AI assistants with email capabilities through AgentMail integration for OTP and notification workflows
Related MCP Servers
View moremarkitdown
Python tool for converting files and office documents to Markdown.
firecrawl
The context API to search, scrape, and interact with the web at scale. 🔥
prompts.chat
f.k.a. Awesome ChatGPT Prompts. Share, discover, and collect prompts from the community. Free and open source — self-host for your organization with complete privacy.
langflow
Langflow is a powerful tool for building and deploying AI-powered agents and workflows.
dothething — FAQ
What is the dothething MCP server?+
Dothething functions as an MCP server when launched with the --browsermcp flag, exposing browser automation, web search, and page fetching tools to other AI agents via the stdio protocol. It can also run as a standalone autonomous agent.
How do I connect dothething to Claude Desktop or another MCP client?+
Add dothething to your MCP client's configuration JSON with the command pointing to the installed dtt script and the --browsermcp flag. The server provides tools like dtt_browser, dtt_fetch, dtt_browser_agent, and dtt_browser_session.
What API keys does dothething require?+
OpenRouter API key is required for the browser agent tool. Optional keys include 2Captcha for automated captcha solving, Serper for enhanced search, and AgentMail for email capabilities. MCP mode only needs OpenRouter when using dtt_browser_agent; search and fetch tools work without it.
Which AI clients work with the dothething MCP server?+
Any MCP-compatible client can connect to dothething when run in --browsermcp mode, including Claude Desktop and Claude Code. The server follows the standard stdio MCP protocol.
Is dothething free to use?+
The dothething software is open source under BSD 3-Clause license, but requires an OpenRouter API key for model calls, which incurs usage costs. Optional services like 2Captcha also charge per use.
How do browser sessions work across multiple runs?+
Named browser sessions persist Firefox profiles with cookies and login state in ~/.dtt/browser-sessions/. Use --browser-session work or the session parameter in MCP tools to reuse logins across separate tasks without re-authenticating.
How do I install dothething?+
Open the source repository on GitHub and follow its README. dothething is a mcp server — MCP Agents Market links you directly to the official repo.
Is dothething free?+
dothething is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.