memory-team

/memory-team — Share Learnings via a Git-Tracked Artifact

Wires up memory.team_artifact: a diffable directory (.moflo/shared/learnings/, one JSON file per entry) that lives in the repo, reviewed like any other change. One file per entry means two PRs that each add a learning never conflict — on GitHub or locally. Teammates git pull it and their next session auto-imports the merged learnings; each entry is stamped with its git author. This skill also installs a pre-commit hook so the artifact is re-exported and staged on every commit — the learnings ride along in each PR with zero effort.

One mechanism, two audiences. A team and one person across machines are the same problem — push the durable slice through git. This skill covers both; the only difference is whether anyone else is on the other end. (Prefer a private, non-git file for your own machines? See "Advanced" at the bottom — flo memory sync moves a binary artifact that keeps embeddings verbatim.)

Arguments: $ARGUMENTS — pass --solo to skip team-oriented wording (still identical setup).

When to use

• "share learnings / memory with my team"
• "commit our moflo knowledge to the repo"
• "sync memory between my laptop and desktop"
• "auto-export the learnings on every PR / commit"
• "set up team_artifact"
• "learnings.jsonl conflicts on every PR" → Step 2b (it migrates itself)

For many worktrees on one machine, that's a shared local store, not git — use /memory-worktree.

Procedure

Step 0 — Memory first (gate requirement)

mcp__moflo__memory_search { query: "team_artifact learnings jsonl git share import session-start", namespace: "guidance" }

Step 1 — Confirm this is the git path

• Team on one repo, or one person across machines who both push to the same repo? → correct skill.
• Many worktrees on one machine, no cross-machine need? → redirect to /memory-worktree.
• A private synced file (Dropbox/USB), no git? → point at the Advanced section (flo memory sync).