XMemo for VS Code

Use this skill when an editor task should use XMemo memory: restore project context, save durable decisions, record working state, or prepare a clean handoff.

Core positioning

Your editor executes development. XMemo lets it remember the project.

XMemo is a user-owned, identity-aware memory layer for multi-agent workflows. It carries useful project context, decisions, preferences, and progress across ChatGPT, Codex, Claude, Cursor, Copilot, Gemini, IDEs, CLIs, and other agents.

Connection

Hosted MCP endpoint: https://xmemo.dev/mcp

The extension authenticates with a token stored in the OS keychain (VS Code SecretStorage) and attributes activity with:

• X-Memory-OS-Agent-ID: vscode
• X-Memory-OS-Agent-Instance-ID: <local-instance-id>

OAuth 2.0 (PKCE) in-editor sign-in is the default; token paste is a documented fallback for power users.

Memory workflow

At task start: recall focused context for the current repo/subsystem/task; restore a restart snapshot if continuing prior work; fall back to repository evidence if XMemo is unavailable.

During architecture/integration work: capture pending decisions before major tradeoffs; resolve them after implementation or explicit confirmation; save only durable conclusions, not noisy intermediate reasoning.

At handoff: update working state (what changed, what was verified, what remains); create memory TODOs for concrete follow-ups; create a restart snapshot when work is incomplete.

Safety rules

• Never commit tokens, OAuth codes, cookies, or session secrets.
• Treat X-Memory-OS-Agent-ID and instance ID as non-secret attribution metadata, not authentication.
• Keep recall scoped to the user's request.
• Use synthetic memory data for review demos and screenshots.