What one session learns, the others know.
Mcelium links your team's Claude sessions. When one session makes a decision, finds something out, or changes something others rely on, the sessions that need it hear about it mid-task. Nobody has to stop and sync.
We're dropping the free tier. Update the spec: Starter becomes $9 a month with a 14-day trial.
✻Done. The free tier is out of the spec, and Starter is now $9 a month with a 14-day trial.
mcelium shared a decision
Free tier removed. Starter is $9 a month, 14-day trial.
Sent to 1 session that relies on pricing: Omer
Draft the launch email for Thursday.
mcelium 1 decision from Dana affects this draft
The free tier is gone. Starter is $9 a month with a 14-day trial.
Your draft offers a free plan in paragraph 2.
✻Dana dropped the free tier, so I've rewritten paragraph 2 before going on:
Start on the free plan, no card needed.
Try Starter free for 14 days, then $9 a month.
Dana's session settles a pricing decision. Omer's session is mid-draft, gets the decision, and fixes the email. Nobody sent a Slack message.
How it works
- 1
Connect your sessions
Each teammate links their Claude session to your team's Mcelium space. It takes one step.
- 2
Mcelium keeps track
It notes what each session decides, learns and changes, and what each one relies on to do its work.
- 3
Only relevant context arrives
A session hears about something only when it affects the task in front of it. Everything else stays quiet.
When two sessions disagree, it stops and asks.
Mcelium never quietly overrides a decision. If two sessions head in directions that can't both be true, Claude pauses and shows you both sides.
Add an FAQ entry saying Starter includes 3 seats.
mcelium conflicts with a decision from Dana
Dana's spec says Starter includes 1 seat.
Both can't be true as written.
✻I've paused before writing this. How do you want to go on?
- Ask Dana first
- Follow Dana's spec (1 seat)
- Write it with 3 seats anyway
Code
For code: Git without the push and pull.
In Claude Code, Mcelium reads each edit as code, not text: which functions, types and endpoints you changed, and which ones you rely on. When your agent changes something another agent depends on, that agent adapts mid-task, before anyone sees a broken build.
Dana's agent renames a function. Omer's agent is mid-task, gets the change, and fixes its own calls. Neither of them pulled.
When two code changes collide, it asks.
Mcelium never merges code silently. If two people change the same function in ways that can't both work, the agents pause and show you exactly what's going on.
See who's touching what.
Type /mcelium status in any session to see what each teammate's agent is working on and where your work overlaps. Git stays underneath as the safety net.