manage-memory-capture

Capture session context into durable Markdown files and a structured on-disk index to track decisions, modified files, and pending tasks for future contributors. It writes entries into .workflow/memory/ and updates memory-index.json for fast lookup and programmatic consumption.

Checkpointing a multi-agent phase requires compressing full-session context into a structured Markdown summary with key decisions, execution plans, and pending work. The captured memory entry enables the Commander or future sessions to resume with a concise context overview.

Quick timestamped notes with tags are generated using a dedicated tip mode for lightweight reminders or actionable tips. This creates a quick note file and appends its metadata to memory-index.json, applying parsed tags for easy retrieval and tracking.

Workspace initialization is required and validated before writing session memory files into .workflow/memory/. This ensures the directory structure exists so the Skill can generate deterministic MEM- or TIP- IDs and update memory-index.json without errors.

Indexing session memories appends metadata to memory-index.json, enabling programmatic consumption and fast lookup in AI orchestration workflows. Generated entries receive deterministic MEM- or TIP- IDs, allowing downstream agents to query past decisions and pending tasks by tags or ID.