Mindpm – persistent project/task memory for AI coding assistants (MCP)
MCP-native persistent memory prevents re-explaining projects to Claude every session.

Polished AI assistant, but MCP integration and memory are table stakes in 2025.
Knowledge workers wanting automated task management
Cursor · Zapier Central · Lindy
Our users are currently using Orchid in 3 main ways:
1. Automations: Orchid can integrate with your tools via MCP and run tasks however often you need to. some cool examples we've seen include: daily email & calendar digests, NBA/F1/Soccer live updates, linear tasks reminders, and email organizing.
2. One off tasks: Orchid can write your emails, coordinate meetings, book your trips, track your calories via images you send it, and even manage your socials. Since Orchid has built in memory, it knows how you like to do things, making completing tasks like this quite easy.
3. Reminders & deadlines: Orchid can pick up on reminders you give it or deadlines it picked up via your email, making things slip through the cracks much less often.
We hope Orchid can give you value and assist you in some way in your life/work.
If you have any questions about Orchid, please reach out! We'd also love your feedback if you've given it a try :)
MCP-native persistent memory prevents re-explaining projects to Claude every session.
One `npx` command, no Docker, but OpenClaw and Claude Desktop already solve this better.
Telegram-native agent with memory and HTML apps, but setup is rough alpha.
This is an ambitious, integration-first personal assistant: the repo shows a web/desktop app, CI, Docker infra and a focus on automating Gmail, Calendar, Slack and todos rather than just chat. The project looks product-ready and opinionated about self-hosting and extensibility, but it's operating in an overcrowded category and the PolyForm noncommercial license will limit commercial adoption.
Turns an existing Claude/OpenAI key into something that actually keeps working for you: heartbeat pulses propose tasks, persistent research notes and daily journals build context over time, and browser control plus messaging (Telegram/WhatsApp/Slack) make it proactively useful. Clever local-first architecture and a polished desktop UI give it real stickiness, but it's boxed in by macOS-only delivery and the need for paid third-party models.
MCP server for agent task management is genuinely useful for dev workflows.