Your team shares a codebase.
Now your AI shares a memory.

MemBridge is a menu-bar app and CLI for teams that code with AI. A tiny local daemon watches the session logs your tools already write, distills each one into a small per-project memory, and injects it into the context files every agent reads at startup.

curl -fsSL https://membridge.me/install.sh | sh

_{local-first · no account · no API keys}

The one-liner is macOS (Apple Silicon). Everywhere else: npm install -g @membridgeai/membridge && membridge start. Every install option, the full CLI, configuration, and the FAQ live in the guide — and this page has an animated twin at docs/readme.html (membridge.me).

One session ends. Every agent remembers.

When Andrew's Codex refactors checkout validation, a redacted digest syncs to the team. Next time you open the project, your Claude Code already knows.

Three quiet steps

01 · Watch — Session logs	02 · Distill — Per-project memory	03 · Inject — Context files
A tiny local daemon tails the JSONL session logs your tools already write — Claude Code, Codex, Gemini CLI, any JSONL logger.	Each session is distilled into a small per-project memory: decisions made, gotchas found, who owns what.	The memory lands between markers in CLAUDE.md, AGENTS.md and GEMINI.md — read by every tool at startup.

A memory block, between markers

MemBridge owns only what sits between its markers. The rest of your file is never touched.

<!-- membridge:begin -->
## Project memory · updated 2h ago
checkout retries cap at 3 with exponential backoff (validate.ts)
payment webhooks replay from stripe-cli in dev — never mocked
e2e tests need POSTGRES_ISOLATION=strict or they flake              ← new
andrew owns the pricing service — ask before touching rate tables   ← new
<!-- membridge:end -->

Small surface, sharp edges

• Team feed — a redacted digest syncs to your team — optionally. Everything starts local: no cloud, no account, no API keys until you join one.
• Summaries by the agent itself — session summaries are written by the agent that did the work, not reconstructed by a heuristic after the fact.
• Copy-for-AI digest — one click copies a compact project digest for any chat window that can't read your files.
• MCP server — project memory is also exposed over MCP, for tools that would rather query than read a file.
• Custom adapters — point MemBridge at any tool that logs JSONL with a small adapter. If it writes sessions, it can share them.
• Secret redaction — keys, tokens and private paths are scrubbed before anything leaves your machine.

There's more in the guide: the dashboard tour with screenshots, session summaries and the Stop hook, team sync and privacy, self-hosting, BYOK roadmaps, the CLI table, custom adapter config, and development docs.

---

_{source-available · free core, zero runtime dependencies · membridge.me}