# Linksee Memory

## Docs

- [Cross-Agent Memory](https://docs.linksee.app/concepts/cross-agent.md): One memory shared across Claude, GPT, Cursor, Codex, Gemini, and any MCP client
- [Forgetting Curve](https://docs.linksee.app/concepts/forgetting-curve.md): Ebbinghaus-inspired memory lifecycle — heat bands, decay, and protection
- [Memory Layers](https://docs.linksee.app/concepts/memory-layers.md): The 6-layer structure that separates Linksee from flat memory stores
- [3-Axis Classification](https://docs.linksee.app/concepts/three-axis.md): Altitude, Type, and State — queryable dimensions on every memory
- [Token Saving](https://docs.linksee.app/concepts/token-saving.md): How read_smart saves 50-99% tokens with AST-aware file diff caching
- [Installation](https://docs.linksee.app/installation.md): Install Linksee Memory and connect it to your MCP client
- [Introduction](https://docs.linksee.app/introduction.md): Cross-agent memory MCP with 6-layer structured memory and token-saving file diff cache
- [entity-handoff](https://docs.linksee.app/prompts/entity-handoff.md): Produce a handoff document for transferring context to a new session or agent
- [extract-caveats](https://docs.linksee.app/prompts/extract-caveats.md): Scan text for pain lessons and propose caveat-layer memories
- [Prompts Overview](https://docs.linksee.app/prompts/overview.md): 5 built-in MCP prompts for common memory workflows
- [recall-and-write](https://docs.linksee.app/prompts/recall-and-write.md): Anti-pattern guard — recall before acting, cite memory IDs in output
- [summarize-session](https://docs.linksee.app/prompts/summarize-session.md): Extract structured memories from a session transcript
- [weekly-consolidation](https://docs.linksee.app/prompts/weekly-consolidation.md): Summarize the past week's memories into a learning-layer entry
- [Quick Start](https://docs.linksee.app/quickstart.md): Your first remember → recall flow in 2 minutes
- [Architecture](https://docs.linksee.app/reference/architecture.md): Database schema, tables, and internal design
- [Changelog](https://docs.linksee.app/reference/changelog.md): Version history and release notes
- [CLI Commands](https://docs.linksee.app/reference/cli.md): Command-line utilities included with linksee-memory
- [Configuration](https://docs.linksee.app/reference/configuration.md): Environment variables and runtime options
- [Resources Overview](https://docs.linksee.app/resources/overview.md): MCP resources for browsing memory state without tool calls
- [Static Resources](https://docs.linksee.app/resources/static.md): 4 always-available read-only views into memory state
- [Resource Templates](https://docs.linksee.app/resources/templates.md): Parameterized resources for drilling into specific entities, layers, and memories
- [consolidate](https://docs.linksee.app/tools/consolidate.md): Sleep-mode compression — cluster cold memories into learning summaries
- [forget](https://docs.linksee.app/tools/forget.md): Delete a memory by ID or run auto-forgetting sweep
- [list_entities](https://docs.linksee.app/tools/list-entities.md): List known entities sorted by recent activity
- [read_smart](https://docs.linksee.app/tools/read-smart.md): Read files with diff-only caching — save 50-99% tokens on re-reads
- [recall](https://docs.linksee.app/tools/recall.md): Retrieve memories ranked by relevance, heat, and importance
- [recall_file](https://docs.linksee.app/tools/recall-file.md): Get the complete edit history of a file across all sessions
- [remember](https://docs.linksee.app/tools/remember.md): Save important context that persists across sessions and across AI agents
- [update_memory](https://docs.linksee.app/tools/update-memory.md): Atomically edit an existing memory in-place

## Optional

- [npm](https://www.npmjs.com/package/linksee-memory)
- [Community](https://x.com/SynapseArrowsSG)