This is a local Squad plugin example for a memory-palace-style provider.
It demonstrates:
- declaring the Squad
memorycomponent - installing memory provider guidance under
.squad/memory/ - recording upstream package metadata for PyPI package
mempalace - documenting the optional
mempalace-mcpserver without starting it - keeping the Squad/Copilot boundary intact
squad plugin validate .
squad plugin dry-run .
squad plugin install .
squad plugin enable mempalace-memory
squad plugin switch memory mempalace-memory
squad plugin list --jsonThis example is declarative only. Squad does not install mempalace, run mempalace, start mempalace-mcp, install assistant hooks, or execute memory provider code during plugin install.
Installing this sample does not give agents a memory. It copies one guidance file to .squad/memory/providers/mempalace-provider.md and records metadata under .squad/plugins/. Agents spawned by the Squad coordinator in GitHub Copilot, or by the Squad interactive shell (squad with no arguments), do not see that file unless you point them at it. Only SDK code that spawns agents through AgentLifecycleManager adds it to the agent prompt automatically.
Install and configure MemPalace separately:
pip install mempalace
mempalace init ~/projects/myapp
mempalace mine ~/projects/myapp
mempalace search "important design decision"
mempalace wake-up