shared team memory instead of manually sending markdown files around
jumpyBrain gives AI coding assistants a shared memory. Let you Claude Code, Codex etc. "remember" important decisions or findings in your teams memory.
Just by typing @jumpyBrain remember that we do not like pinapple on pizza. Company policy!
The whole team (their Claude/Codex) will recall this decision when planning the team event. Automatically.
Install jumpyBrain once. The installer sets up shared memory and adds integrations for detected assistants such as Codex, Claude Code, and Pi.
curl -fsSL https://raw.githubusercontent.com/nikoatwork/jumpyBrain/master/install.sh | bashRestart your coding assistant if it was already open.
Then talk to jumpyBrain from the assistant you already use:
| Assistant | Store something | Recall it later |
|---|---|---|
| Codex | @jumpyBrain remember that we do not like pinapple on pizza. Company policy! |
@jumpyBrain what should we remember when planning the team event? |
| Claude Code | @jumpyBrain remember that we do not like pinapple on pizza. Company policy! |
@jumpyBrain what should we remember when planning the team event? |
| Other, like Pi agent | Use jumpyBrain to remember that we do not like pinapple on pizza. Company policy! |
Use jumpyBrain to recall what we should know before planning the team event. |
That is the whole idea: your team stores important context once, and future agents can pull it back when it matters.
Good memories are durable things the team should not rediscover:
- decisions and company preferences
- project conventions
- solved bugs and gotchas
- handoffs and important findings
Do not store secrets, credentials, or private chat noise.
- Works across different coding assistants.
- Keeps memory in readable Markdown files.
- Helps the team own its memory instead of locking it into one model provider.
- Installation:
docs/install.md - Agent workflows:
docs/agent-workflows.md - Memory format:
docs/memory-format.md - Technical details:
docs/technical.md