Lossless Claude FAQs

Question 1

What is Lossless Claude?

Accepted Answer

Lossless Claude establishes a cross-runtime memory layer for Claude Code sessions. It integrates with the OpenClaw lossless-claw SQLite database, enabling persistent context, recall capabilities, and shared memory with other OpenClaw agents.

Question 2

Does Lossless Claude replace Claude Code's built-in context management?

Accepted Answer

No, Lossless Claude works *beside* Claude Code's native context management. Claude Code's harness still controls its active context window and compaction. Lossless Claude provides external recall and persistence without interfering with this core functionality.

Question 3

How does Lossless Claude enhance Claude Code sessions?

Accepted Answer

It enhances sessions by exposing LCM recall tools (like `lcm_grep` for search, `lcm_describe` for detailed lookup) directly to Claude Code via MCP. It also captures session messages for persistent memory and can bootstrap new sessions with recent relevant activity.

Question 4

How is session data stored and accessed?

Accepted Answer

Session data is stored in a shared SQLite database (defaulting to `~/.openclaw/lcm.db`) with a distinct `cc:` prefix for Claude Code session keys. This allows for cross-runtime access and prevents collisions with data from other OpenClaw agents like Daphne.

Question 5

What tools does Lossless Claude provide for memory recall?

Accepted Answer

It provides three key MCP tools: `lcm_grep` for full-text search across messages and summaries, `lcm_describe` for looking up specific messages or summaries by ID with their lineage, and `lcm_stats` for a database health snapshot and prefix breakdown.

Lossless Claude

Lossless Claude

主要功能

使用案例

主要功能

使用案例