Skip to main content

​
General

Usually 10-60 seconds. The Curator watches for new inbox items and attends to them automatically. If things seem delayed, check that the Curator is running via your dashboard or GET /api/v1/workers/status.
Yes. You can create, edit, move, and delete records via the API. Use PATCH /api/v1/vault/records/{path} to edit a record’s metadata or content, or POST /api/v1/vault/records/{path}/move to rename or retype it. See the Vault API Reference for details.
The record is permanently removed from your vault. Any connections from other records that pointed to the deleted record will become broken links. The Janitor will flag these during its next scan and can repair them automatically.
Check specialist history via the API:
  • Janitor history: GET /api/v1/workers/janitor/history
  • Distiller history: GET /api/v1/workers/distiller/history
You can also browse recently created or updated records in your dashboard vault browser.
Alfred uses state-of-the-art language models optimized for knowledge extraction. You can choose which model each specialist uses from your dashboard’s Services page — including models from Anthropic, OpenAI, Google, xAI, and others via OpenRouter.

​
Intuition

Intuition is Alfred’s learning system. As you route inputs and make organizational decisions, Alfred observes the patterns. Over time, those observations are refined into instincts — learned routing patterns that let Alfred handle routine decisions on your behalf. See the Intuition guide for the full picture.
Alfred needs at least 3 observations of the same pattern before creating an instinct, and new instincts start with high discretion thresholds. In practice, expect Alfred to begin handling its first routine decisions within 1-2 weeks of regular use. The more you route through Alfred, the faster it learns.
Yes. Disable it with POST /api/v1/learning/disable or from your dashboard. Disabling pauses all learning processes but preserves your existing observations and instincts. Re-enable at any time to pick up where you left off.
Route it correctly yourself — this creates a new observation that will be factored into the next nightly reflection. Alfred adjusts its instincts based on new evidence. If an instinct becomes consistently wrong, reflection will deprecate it automatically.

​
Troubleshooting

  1. Check the Curator status — Go to your dashboard Command Center or call GET /api/v1/workers/status
  2. Restart the Curator — Try POST /api/v1/workers/restart
  3. Manual run — Trigger a manual run with POST /api/v1/workers/process
If the issue persists, check your Alfred’s health at GET /api/v1/admin/health.
Up to 10 per account. You can manage your keys in Settings > API Keys in your dashboard. Revoke keys you no longer need to free up slots.
Yes. Your Alfred runs daily automatic encrypted backups. You don’t need to manage backups yourself.
Yes. Your vault is fully Obsidian-compatible. Pair your device to access it directly. See the Device Pairing guide for instructions on connecting Obsidian to your Alfred.
Ask the Distiller:
  1. Scan for extractable knowledge: POST /api/v1/workers/distiller/scan
  2. Run extraction: POST /api/v1/workers/distiller/run
The Distiller will create learning records (assumptions, decisions, constraints, contradictions, syntheses) from your existing vault content.

​
Getting help

If you can’t find the answer here, check: