MCP servers · ancientice

Palace: local-first memory for coding agents (MCP server)

A local MCP server that gives coding agents memory across sessions, stored verbatim in a single SQLite file with local embeddings.

This text was machine-translated from the publisher's own information and has not been checked by a human editor. The publisher's original text is the reference.

This resource was imported automatically from the sources listed below and has not been reviewed by the AI Gate team yet. The description is the publisher's own, as published. Check the repository before installing.

What it does

Palace stores what coding agents do and learn, such as decisions, fixes, commands and preferences, as text kept verbatim in one SQLite file. Items are organised into wings, rooms and drawers, alongside a time-stamped knowledge graph and per-agent diaries. Search combines BM25 keyword scoring with cosine similarity over local embeddings, and each result cites its wing, room, source file and score breakdown. Hooks inject relevant memory at session start, on prompts and during investigation. It is listed for clients including Claude Code, Codex, Cursor, Gemini CLI, VS Code, Cline, Zed, Goose, Windsurf and Claude Desktop.

It suits cases where agents repeat rejected approaches or rediscover conventions in every new session. Topic wings can hold chat history or meeting notes without a project folder, and the creative and personal profiles adapt it to fiction, worldbuilding, coaching or household notes.

Setup means installing the palace binary through a shell script, PowerShell script, Homebrew, the MCP registry bundle or from source, then running palace install to connect agents and restarting the agent app. The default embedding model (about 23 MB) downloads from Hugging Face on first mine or search and is cached locally. Other supported local models can be tried, and migrate rebuilds vectors from the stored text in batches.

Limitations and requirements

Not supported on macOS Intel, as there is no ONNX Runtime build for it. The embedding model downloads from Hugging Face the first time it is used. Remote mode, if enabled, sends data off the machine. Experimental embedding models scored below the default on the publisher's coding-agent eval. Homebrew needs brew update before upgrading, and releases that change the memory protocol require running palace install again.

Safety

Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.

  • Runs code or changes an assistant's behaviour

Sources

  • Official MCP Registry first seen 13 Nov 2026 · last seen 17 Nov 2026 · its page on the source
    Palace

    Local-first memory for coding agents — MCP server, single SQLite file, local embeddings

    v0.20.2
    As shown on registry.modelcontextprotocol.io when we last read it
MCP servers
hashicorp

Terraform MCP Server (HashiCorp)

HashiCorp's MCP server that gives an assistant current Terraform Registry information on providers, modules and policies, and can manage HCP Terraform workspaces.

1.6K GitHub starsCloud & DevOpsSoftware development
Updated 17 Nov 2026
MCP servers
modelcontextprotocol

MCP reference servers

Seven small example servers maintained by the Model Context Protocol project: files, Git, web fetching, memory, time, step-by-step thinking and a test server.

92K GitHub starsSoftware development
Updated 18 Nov 2026
MCP servers
redis

Redis MCP Server

Redis's official MCP server that lets an AI assistant store, read and search data in a Redis database, including streams, JSON documents and vector indexes.

642 GitHub starsData & databasesSoftware development
Updated 17 Nov 2026
MCP servers
firecrawl

Firecrawl MCP Server (web search and scraping)

Firecrawl's MCP server that lets an AI assistant search the web, turn pages and PDFs into clean Markdown, crawl sites and interact with dynamic pages.

7.8K GitHub starsWeb, search & browserProductivity & automation
Updated 17 Nov 2026