AgentMemoryBridge: Shared memory across coding agents

Automatically carry context between Claude Code and Codex so you never restart from scratch

Problem

Developers switch between Claude Code and Codex when one hits its quota, losing all project context, decisions, and TODOs. They have to re-explain everything from zero. Someone is already building plug-and-play AI memory across agents, proving the pain is real and urgent.

Solution

A local-first memory layer that captures key decisions, file changes, and TODOs from any coding agent session, generates a structured handoff context, and injects it into the next session automatically. Works with Claude Code, Codex, and Gemini CLI.

Users

Mid-to-senior indie developers and small teams subscribing to multiple coding agents

MVP

Support session export/import for Claude Code and Codex; auto-generate project summary, decision log, and unfinished TODOs; local SQLite storage, no code upload; CLI commands and slash-command triggers for handoff; a simple switcher dashboard for handoff history.

Revenue

Free tier for single project and single agent switch; Pro subscription for multi-project, multi-agent, and custom summary templates; team plan with shared memory vault.

Why now

Coding agent subscriptions are expensive with tight quotas, making multi-agent rotation the norm, yet cross-tool context migration has no good solution.