Your next AI session should start with what your team already knows.
Lore keeps the decisions, conventions and history your team builds up, and gives every connected agent the part it needs for the task in front of it.
Let the knowledge stay with the project.
Memory belongs to the organization and the project, not to one developer, one IDE or one model. When someone switches tools or joins the team, the project’s knowledge is already there.
Find the decision behind the task.
Search works by words, by meaning and by relationships at once, so the agent finds the decision even when the task describes it differently.
Keep useful knowledge current.
One set of instructions for every agent.
Stop maintaining CLAUDE.md and AGENTS.md in every repository. Write your rules once, layered from the organization down to the team, the project and the individual, and every connected agent receives them at session start. Every edit is recorded, and a check keeps project rules from creeping back into local files.
Governed context is also the fastest context.
For repeated workloads, Lore prepares context ahead of time with versioned recipes, preparation jobs and cache residency, for hosted model providers or your own vLLM and SGLang deployments.
Shape what each client receives.
Capture you can see.
Claude Code, Codex, Cursor, Windsurf, Cline and Gemini CLI, plus any MCP client connected to Lore. The health panel marks each client fresh, stale, broken or absent, so a quiet client never passes for a working one.
Connect in one command.
lore memory connect sets up the Lore MCP server in your AI client: 13 tools up front and 88 more discovered on demand, served over HTTP.
