First Capture

Your first thought is captured, encrypted, and searchable in seconds.

The Capture Window

Press Ctrl+Shift+P (default hotkey). A capture window appears — already focused on the editor.

Three window modes (configurable in Settings → Capture):

Mode	Behavior
Taskbar	Slides up from the bottom, 940×200, undecorated, always-on-top
Center	Centered popup, 500×300, decorated
Window	Full window mode, resizable

Writing Your First Thought

Type naturally. Some examples:

Decided to use Redis for session caching instead of Memcached.
Redis pub/sub gives us real-time invalidation which Memcached can't do.

/idea: Build a CLI tool that auto-generates .cortexa-context.md on git checkout

/debug: The auth middleware was silently swallowing 403 errors because
the error handler checked status.is_success() instead of !status.is_client_error()

Save Options

Before pressing Ctrl+Enter to save, toggle the action pills:

• Summarize — LLM generates a one-line summary
• Deep Think — runs the expand → critique → synthesize pipeline
• Research — web search + LLM synthesis, linked to this thought

Or use a slash template (/idea:, /todo:, /quote:) which auto-applies the template's default actions.

What Happens After Save

1. Encrypted — AES-256-GCM, written to disk as an encrypted blob
2. Indexed — preview + summary added to FTS5 for keyword search
3. Queued for embedding — background worker generates a semantic vector
4. Auto-linked — once embedded, the knowledge graph links it to similar thoughts
5. Auto-shared — if Hive is enabled, pushed to peers in real time
6. Notification — "Related thoughts found" if similar items exist in your vault

Viewing Your Thought

Click the thought in the list to expand it. You'll see:

• Full decrypted content
• Summary (if generated)
• Tags (inline #hashtags auto-extracted + template tags)
• Related items (from the knowledge graph)
• Thesis sections (if Deep Think was triggered)
• Conversation thread (reply to ask follow-ups)
• Edit button (pencil icon) to modify content