Learnings

Learnings are your team’s institutional memory for AI assistants: bugs you fixed, stack quirks, architecture decisions. They live in the private team repo, not in public Core.

Paths

${SHARED_AGENTS_HOME}/team/learnings/pending/YYYY-MM-DD-short-slug.md
${SHARED_AGENTS_HOME}/team/learnings/approved/…
${SHARED_AGENTS_HOME}/team/learnings/index.yaml

Resolve before writing:

sa pending path 2026-06-02-my-slug

Workflow

Agent drafts pending/  →  sa pending push  →  sa review  →  approved/

# Agent (after user approves creating a learning)
# Write to path from: sa pending path <slug>

sa pending push <file>     # commit + push draft to team remote
sa review list
sa review <file>           # promote → approved/ + update index.yaml + push
sa sync                    # teammates pull approved content

Review

sa review                  # interactive file picker
sa review 2026-06-02-my-slug.md
sa review dry <file>       # preview only

Unapprove or fix mistakes

sa unapprove list
sa unapprove <id-or-file>
# interactive: [1] Delete  [2] Move to pending/  [q] Cancel

Flags: --to-pending, --delete, --dry-run, --no-git, -y

Index file

team/learnings/index.yaml catalogs approved learnings (ids, domains, titles). Updated by sa review.

Agents should read the index and search approved/ before non-trivial tasks (after sa sync).

Agent rules

1. Never write to approved/ directly
2. Never put learnings in the Core repo or client project
3. Always use sa pending path or explicit $SHARED_AGENTS_HOME/team/learnings/pending/…
4. Ask the user before creating a draft; use skill capture-learning
5. Only approved learnings count as team knowledge

See Canonical paths.

Checks & migration

sa team verify
sa status                  # pending items needing review

Legacy layout at ~/.shared-agents/learnings/: Migrate team data · sa team migrate

Related

• Team setup
• CLI reference — sa review, sa pending, sa unapprove
• capture-learning skill