vault-groom

Run a health check on the knowledge base and fix issues.

Team Mode (Large Vaults)

For large vaults (500+ entries), spawn a team to parallelize grooming:

team_spawn(
  team_name: "vault-grooming",
  purpose: "Audit, archive, and fix issues across a large vault",
  roles: [
    %{name: "vault-auditor", role: "researcher"},
    %{name: "vault-archiver", role: "coder"},
    %{name: "vault-fixer", role: "coder"}
  ]
)

Team roles:

• vault-auditor: Runs vault_audit(scope: "full") and catalogs all issues. Shares the report with archiver and fixer.
• vault-archiver: Processes completed task archival (vault_kanban(action: "archive")), archives old checkins, and moves stale entries.
• vault-fixer: Applies approved metadata fixes, corrects frontmatter, resolves structural issues.

For smaller vaults, skip team mode and process single-agent.

---

Audit

vault_audit(scope: "full") — get the full quality report.

Check Task Board

vault_kanban(action: "list", filter_column: "done") — find completed tasks ready for archive. vault_kanban(action: "list", filter_column: "in_progress") — check for stale items (no activity in 7+ days).

Check Temporal Entries

vault_list(entry_type: "checkin") — find checkins older than 30 days for archiving.

Present Findings

Format as a structured report:

• Task issues (missing metadata, stale items)
• Completed tasks ready to archive
• Old checkins ready to archive
• Quality issues from vault_audit
• OKR/milestone status inconsistencies

Interactive Fix

Use ask_user to confirm each category of fix:

• "Archive these X completed tasks?"
• "Archive these Y old checkins?"
• "Fix these Z metadata issues?"

Apply Fixes

• Archive tasks: vault_kanban(action: "archive")
• Archive checkins: vault_update_entry(status: "archived") for each
• Fix metadata: vault_update_entry(frontmatter_updates: "...") for each

Refresh

vault_dashboard(dashboard_type: "index") after all fixes.