vshulcz/deja-vu

Search your past AI coding sessions — Claude Code, Codex, Cursor and 17 more. Indexes the session history they already wrote to disk, including months from before you installed it, and recalls it in any of them. No LLM, no embeddings, one local Go binary.

What it solves

Coding agents often lack long-term memory, causing them to re-solve the same bugs or forget decisions made in previous sessions. deja-vu provides a unified memory layer that indexes the local history of multiple coding agents (such as Claude Code, Cursor, and Codex), allowing any agent to recall previous solutions, commands, and decisions across different sessions and tools.

How it works

The project creates a local inverted index of agent session files (JSONL and SQLite stores). It parses these logs, redacts sensitive credentials like API keys and tokens, and stores them in a cache. It then integrates with agents via the Model Context Protocol (MCP), providing tools like recall and blame that agents can call to retrieve relevant historical context. It also supports optional semantic search using local or remote embedding endpoints (e.g., Ollama or OpenAI).

Who it’s for

Developers who use multiple AI coding assistants and want their agents to have a persistent, cross-agent memory of their project's history and technical decisions.

Highlights

  • Cross-Agent Recall: A solution found in one agent is accessible to any other supported agent.
  • Retroactive Indexing: Indexes existing history from before the tool was installed.
  • Action-Point Recall: Can trigger recall automatically before an agent edits a file or runs a command via hooks.
  • Privacy-First: All indexing and search are local; credentials are automatically redacted during indexing.
  • Broad Support: Compatible with over 20 coding agent harnesses, including Claude Code, Cursor, and Zed.
  • High Performance: Millisecond lookups over gigabytes of history with a small index footprint.

Related

  • Project
    lovstudio/Ataru

    A local search engine for AI conversation history that indexes local transcripts from agents like Claude and Codex to make past AI work retrievable.

  • Project
    dezgit2025/auto-memory

    A zero-dependency Python CLI that allows AI coding agents to recall previous session history and modified files from Copilot CLI, Claude Code, and IDEs to prevent context loss.

  • Project
    kitfunso/hippo-memory

    A biological-inspired memory layer for AI agents that uses decay and reinforcement to ensure agents remember important lessons and forget trivial information across different tools.

  • Project
    JuliusBrussee/cavemem

    A cross-agent persistent memory system for coding assistants that uses compressed local storage and MCP to allow AI agents to remember information across sessions.

  • Project
    akitaonrails/ai-memory

    A long-term memory system for AI coding agents that creates a shared, persistent Markdown wiki from session observations to enable seamless context handoffs between different agents.