from agent-native-os10
Capture loose inputs (links, notes, voice transcripts) into a durable second-brain: raw-first staging, one-question follow-ups, and handoff to Cairns L2/L1 flow
Second Brain Capture provides a lightweight, opinionated flow to turn ephemeral user inputs into durable, searchable memory. It listens for user cues like 'remember this' or 'save this', for pasted links, scratchpad items, or transcript text, and creates a raw timestamped capture file before any summarization. The emphasis is 'raw-first' — preserve the original content, classify it, and then enrich only when useful (for example, fetching/ summarizing linked pages). The skill also defines when to escalate captures into higher-quality Cairns L2 cards or short L1 waypoints.
Use this skill when the user explicitly asks to save or remember something, drops a rough thought or note, shares a URL or transcript, or makes a decision-like statement that should be recorded. Activate for Slack scratchpad items, voice memo transcripts, screenshots, or any input the user says should be durable. Only enrich links or claims that benefit from extra context.
This is platform-agnostic procedural guidance suitable for Claude Code, Copilot-style agents, or any assistant that can read/write files and call web fetch tools. It is primarily a workflow spec rather than an executable toolset.
This skill has not been reviewed by our automated audit pipeline yet.
