mcpField
M
Mindswap

Mindswap

Local
github.com/ShipOrBleedMemory & Knowledge7Updated 25w ago

About Mindswap

Mindswap is a local-first context and memory server for developers and coding agents, designed to carry project state and decisions across sessions. Its CLI saves and searches memory, creates handoff briefings, and can share context with AI clients through MCP.

Key features

  1. 01Save tasks, Git state, and decisions
  2. 02Resume work with handoff briefings
  3. 03Search memory with citations
  4. 04Manage blockers, assumptions, and questions
  5. 05Separate project and personal memory scopes
  6. 06Sync state across machines

Use cases

  1. 01Resume a coding task in a later session
  2. 02Find the rationale behind a project decision
  3. 03Carry personal preferences across projects
  4. 04Share project context with AI clients

MCP support

Mindswap exposes project and personal context to AI clients through MCP commands, including mcp and mcp-http. It also provides an MCP installation command.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"mindswap": {
"command": "npx",
"args": [
"-y",
"mindswap"
]
}
}
}
Run commandnpx -y mindswap

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npmmindswap3.3.0

Tools

4

Saved tool list · updated 2w ago

ParameterTypeRequiredDescription
focusenumNoWhat to focus on: "all" for full context, "task" for current work only, "decisions" for architecture decisions, "recent" for latest changesall | task | decisions | recent
compactbooleanNoReturn token-optimized compact format (fewer tokens, same info)

Details

From the project README

mindswap

npm version License: MIT

Keep project context and personal AI memory local so tools can continue work without re-explaining the same context.

Why it exists

AI sessions reset too often. mindswap saves the current task, decisions, blockers, and handoff context so the next tool can pick up cleanly.

Install

npm install mindswap --save-dev

Quick start

npx mindswap init
npx mindswap
npx mindswap doctor
npx mindswap resume
npx mindswap ask "Why did we choose JWT?"

What it gives you

  • init to set up a repo and import existing AI context
  • save to capture the current task, git state, and decisions
  • doctor to check setup health and stale context
  • resume to start with a clean briefing
  • ask to search project memory with citations
  • memory to manage blockers, assumptions, questions, and resolutions
  • --global memory and ask scope for personal cross-tool memory under ~/.mindswap/
  • reindex to rebuild the local SQLite search index from your file-based memory
  • sync to share continuity state across machines
  • mcp and mcp-http to expose the same context to AI clients

Global personal memory

MindSwap now supports two local memory scopes:

  • repo memory in <repo>/.mindswap/
  • personal memory in ~/.mindswap/

Use global memory when a preference or learning should follow you across projects and tools.

npx mindswap log "Prefer concise explanations" --type assumption --global
npx mindswap memory list --scope all
npx mindswap ask "What explanation style should we use?" --scope all
npx mindswap reindex --scope all

MCP and AI tools

mindswap generates context for tools like Claude Code, Cursor, Copilot, Codex, Windsurf, Cline, Roo, Aider, Amp, Gemini CLI, and AGENTS.md-based workflows.

npx mindswap mcp-install
npx mindswap mcp-http

Project state

The main repo data lives in .mindswap/:

.mindswap/
├── HANDOFF.md
├── state.json
├── decisions.log
├── memory.json
├── config.json
├── branches/
└── history/

npm package

Use it

If mindswap helps your workflow, star the repo and keep improving the handoff loop.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.