aget-capture-observation

To capture research observations safely, append timestamped, typed entries to session files or a scratchpad. This preserves transient signals like vocabulary candidates and web findings while enforcing constraints that prevent direct edits to formal specs or vocabulary files.

Preserving transient research signals involves classifying them by type and appending ISO 8601 timestamped entries to a knowledge base. This prevents premature commits to formal specifications and ensures valuable patterns or process gaps are retained for later review.

Yes, autonomous scanning can surface valuable web findings and patterns. When invoked without arguments, the system scans recent session history and prioritizes the top 3-5 signals based on novelty and governance value for immediate capture.

Time-budgeted capture modes scale classification and formalization effort to your allotted time. Shallow-to-deep modes ranging from 2 to 30+ minutes allow you to record research observations with appropriate linking and typing based on available time.

Yes, capturing web findings requires ISO 8601 timestamps and source attribution. These governance-safe constraints ensure that all web findings appended to session files or scratchpads maintain traceability and comply with knowledge base standards.

You should not use a scratchpad if you intend to directly modify formal specs, vocabulary files, or backlog artifacts. The capture mechanism enforces inviolable constraints against direct edits, meaning scratchpads are strictly for preserving observations, not committing changes.