Path-Native Memory for Agents
Lore gives every memory a durable path, then recalls the right context across sessions and runtimes.
# Works everywhere. Installs Lore and all supported plugins.
$ curl -fsSL https://raw.githubusercontent.com/FFatTiger/lore/main/scripts/install.sh | bashConnects toClaude Code Codex πPi OpenClaw Hermes Generic MCP
Awesome Features
All the memory lifecycle pieces in one UI.
Boot baseline
Every session starts with stable identity, workflow, user, and runtime memories.
Recall before reply
Relevant memory candidates are injected before the agent replies.
URI-first graph
Memories live at durable addresses like core://agent and project://lore.
Read before trust
Agents open recalled nodes before relying on them, then write with cautious tools.
Dream maintenance
Dream review keeps stale nodes, glossary, disclosure, and history under control.
Runtime bridge
Claude Code, Codex, Pi, OpenClaw, Hermes, and MCP clients share one layer.
Product screenshots
Four views that explain how Lore works.
Real product views from the memory browser, recall analytics, recall detail, and dream maintenance workflow.
Memory home
A home base for every memory.
Browse durable memory paths, priorities, disclosures, glossary terms, and node history from one clean graph surface.
Recall analysis
See why context was selected.
Inspect why a memory was recalled, compare score signals, and verify exactly what context reached the agent.
Recall detail
Debug a single recall decision.
Open one query to trace candidate URIs, score cues, stage counts, and the final memories selected for injection.
Dream review
Turn memory upkeep into a review flow.
Review maintenance suggestions before Lore merges stale notes, cleans glossary drift, or rewrites memory nodes.
