# MCP Lance Context > Build or update the lance-context MCP server and Claude Code example that uses lance-context as a memory/knowledge store. Use when this capability is needed. - Skill: `tomevault-io/mcp-lance-context` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/mcp-lance-context` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/mcp-lance-context/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/mcp-lance-context --- # MCP Lance Context ## Overview Use this skill when adding or updating the MCP server for lance-context or the Claude Code example that demonstrates memory/knowledge workflows. ## Files to touch - `examples/mcp-claude-code/server/lance_context_mcp.py` - `examples/mcp-claude-code/README.md` - `examples/mcp-claude-code/requirements.txt` ## Workflow 1. Keep MCP tool outputs JSON-serializable (convert datetimes to strings). 2. Add or adjust tools that write to `Context.add()` and read via `Context.list()`. 3. Update the README instructions and verify CLI commands stay accurate. 4. If dependencies change, update `requirements.txt`. ## Claude Code setup Use project-scoped MCP configuration with `claude mcp add --transport stdio --scope project`. This writes a `.mcp.json` file in the project directory. --- > Source: [lance-format/lance-context](https://github.com/lance-format/lance-context) — distributed by [TomeVault](https://tomevault.io).