Team Sync

Share captured knowledge across your team through a Cloudflare-backed sync layer. Each teammate keeps their knowledge local, and Team Sync mirrors your team’s knowledge into a shared, queryable store — so everyone’s agents draw on the same spores, sessions, plans, and graph without anyone having to think about it.

Team Sync also exposes a secure Cloud MCP Server so cloud agents (Anthropic Managed Agents, n8n, and the like) can read your team’s knowledge. See Cloud MCP for that side.

What you get

• New spores, sessions, plans, and graph edges sync automatically in the background.
• Search fans out across local and team knowledge and merges by relevance, tagged with its source.
• Team results are additive — if the team store is slow or unreachable, your local results still return.
• Assigning a project to a team backfills its existing knowledge once; everything after syncs as you work.
• A team roster shows who’s connected.
• Runs on the Cloudflare free tier for small teams.

Your local data stays the source of truth — the team store is a queryable mirror, and nothing is pulled back down to overwrite it. Every record is attributed to the machine that created it.

Set up a team

There are two roles: one person provisions the team’s cloud infrastructure, and everyone else joins with the details that person shares. Only the person provisioning needs Wrangler and the operator CLI — teammates join straight from the dashboard with nothing extra to install.

1. Provision the team (one person, once)

Install Wrangler and the operator CLI, sign in to Cloudflare, then create the team with a name:

npm install -g wrangler @goondocks/myco-team
wrangler login
myco-team install --name "Acme Core"

This deploys the team’s sync Worker and registers the team on your machine. To serve it from your own domain, add --domain <your-zone>. When it finishes, it prints the team’s Worker URL and Team key — share both with your teammates through a private channel. The Team key authorizes writes to the team, so keep it secret.

2. Teammates join (everyone, from the dashboard)

Each teammate opens the Team page in their Myco dashboard (http://localhost:20915/team, or run myco open), goes to the Teams tab, and under Join a team pastes the Worker URL and Team key. That registers the team on their machine — no Wrangler, no operator CLI, nothing else to install.

3. Choose which projects sync

On the Teams tab, assign each project to a team. A project syncs to exactly one team; leave it unassigned to keep it local. As soon as you assign a project, its existing knowledge backfills to the team store and new knowledge syncs as you work.

Syncing is durable: changes are queued and retried, and anything stuck shows up on the Sync tab, where you can retry or discard it.

What syncs

Teammates see what was asked and answered — not every file read or shell command. That keeps the team store useful and the cost bounded.

The Team page

A team selector in the top-right scopes what you’re viewing to one of your teams. The page has four tabs:

• Teams — machine-wide. Join or add a team, see every team registered on this machine, and assign each project to a team. Teams you provisioned also show their one-line update command; teams you only joined show a Leave team action.
• Status — the selected team’s connection health, the Worker URL and Team key to hand to a new teammate, this machine’s identity in the team, and the team’s Cloud MCP endpoint and token (with a config snippet for cloud agents and a rotate action).
• Sync — the selected team’s sync health: how much is left to send, failed items with Retry and Discard, and one-click backfill.
• Members — the selected team’s roster: every machine connected to the team, with yours marked.

Leaving a team

To stop syncing to a team, open the Teams tab and choose Leave team. That removes the team from this machine and clears its pending queue — it does not touch the team’s cloud Worker or your local data, and other members are unaffected. (To tear the cloud infrastructure down entirely, the operator runs myco-team destroy --team-id <id>.)

Machine identity

Every synced record is tagged with a machine identity — a deterministic {github_username}_{machine_hash} (e.g. chris_a7b3c2). This attributes knowledge to its source and lets you tell “my data” from “team data”. It’s generated once per machine and cached at .myco/machine_id. Nothing to configure.

Search fan-out

When a project syncs to a team, search hits your local store and the team store in parallel, then merges by relevance. Each result is tagged so you can see where it came from:

• source: "local" — from this machine
• source: "team:chris_a7b3c2" — from the team store, attributed

If the team store is unreachable within a short timeout, local results return on their own. Team search is always additive, never blocking.

Cloud embedding alternative

If you’d rather not run Ollama locally for embeddings, point at Cloudflare Workers AI — it uses the same model as the Worker, so embeddings are directly comparable:

# myco.yaml
embedding:
  provider: openai-compatible
  model: "@cf/baai/bge-m3"
  base_url: https://api.cloudflare.com/client/v4/accounts/{account_id}/ai/v1

Store your Cloudflare API token in secrets.env.

Backup & restore

Independent of Team Sync, Myco creates local backups for resilience. Configure the backup directory on the Operations page, or click Backup Now for an on-demand backup. Restore supports a dry-run preview, and cross-machine restore preserves attribution, so you can pull a teammate’s backup without losing who said what.

Backups include all knowledge but exclude logs, tool-call activity, and vector embeddings (rebuilt automatically after restore).

Managing the Worker (operators)

The person who provisioned a team manages its Worker. Each team you provisioned shows its update command on the Teams tab — copy and run it to redeploy the Worker at your installed Myco version:

myco-team update --team-id <team-id>

The other operator commands take the same --team-id: status, rotate-tokens, reindex-vectors, and destroy. To update the operator CLI itself:

npm update -g @goondocks/myco-team

The Worker keeps the cloud side lightweight: your local data stays the source of truth, the Worker holds a synced mirror for team search and Cloud MCP, and syncing is queued and retryable, so a transient network or Cloudflare hiccup never blocks local work.

Cost

A small team (2–5 developers) stays comfortably within the Cloudflare free tier. The $5/month paid tier provides significant headroom if your team outgrows it.