Skip to main content
MCP server CX Codex

Lians Agent Memory — Claude Code MCP server: what it does, install

This MCP server tracks task state across agent sessions and rejects unverified completion. It stops models from reporting tasks as done without real test or CI confirmation.

Install into Claude Code · Platform CLI All methods and platforms ↓
claude mcp add --env LIANS_URL=<LIANS_URL> --env LIANS_API_KEY=<LIANS_API_KEY> --env LIANS_AGENT_ID=<LIANS_AGENT_ID> --env LIANS_LOCAL_DB=<LIANS_LOCAL_DB> --env LIANS_NAMESPACE=<LIANS_NAMESPACE> --transport stdio lians-agent-memory -- uvx lians-sdk
Open on GitHub

What Lians Agent Memory does

Lians keeps track of decisions and open tasks across terminal runs or IDE prompts. Instead of replaying full chat logs, it extracts changed items, current constraints, and next steps. Checkpoints bind directly to the repository state, stopping the assistant from relying on outdated findings.

The server enforces a separated trust model. Self-reported assertions like "tests passed" remain unverified context and cannot open the handoff gate. Progress toward human review requires measurable evidence from local test executions, matched CI commit checks, or direct operator approval.

Standard recovery runs locally without requiring a remote account or vendor API token. By default, records go to SQLite at ~/.lians/mcp.db. Running the server via uvx requires uv on your machine.

What it helps with

  • resume interrupted tasks across fresh agent sessions
  • block task completion when claims lack test evidence
  • invalidate stale checkpoints after repository changes
  • hand off structured project context instead of full chat histories

Example prompts for the agent

  • Recall the architectural decisions made for the authentication flow.
  • Save this database migration checkpoint and show open checklist items.
  • Prepare the current task state for human review.

How to install Lians Agent Memory

Platform CLI

claude mcp add --env LIANS_URL=<LIANS_URL> --env LIANS_API_KEY=<LIANS_API_KEY> --env LIANS_AGENT_ID=<LIANS_AGENT_ID> --env LIANS_LOCAL_DB=<LIANS_LOCAL_DB> --env LIANS_NAMESPACE=<LIANS_NAMESPACE> --transport stdio lians-agent-memory -- uvx lians-sdk

Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from lians-sdk <command>

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about Lians Agent Memory

How to install Lians Agent Memory in Claude Code?

Run claude mcp add lians -e LIANS_API_KEY=<LIANS_API_KEY> -- uvx lians-sdk. If you need custom environment variables, run `claude mcp add --env LIANS_URL=<LIANS_URL> --env LIANS_API_KEY=<LIANS_API_KEY> --env LIANS_AGENT_ID=<LIANS_AGENT_ID> --env LIANS_LOCAL_DB=<LIANS_LOCAL_DB> --env LIANS_NAMESPACE=<LIANS_NAMESPACE> --transport stdio lians-agent-memory -- uvx lians-sdk`.

What does Lians Agent Memory do?

It maintains persistent task state across coding sessions and prevents assistants from marking jobs finished without measured evidence.

Where does the server save data locally?

By default, the server writes local records to a database at ~/.lians/mcp.db.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor