waypointset

name	waypoint:set
description	Save session context to vector memory for seamless resumption
user-invocable	true
disable-model-invocation	true

Create a comprehensive waypoint snapshot using vector-memory.

1. Extract & Store Memories

Review the session for significant items worth persisting long-term.

Extract:

• Architectural decisions and rationale
• Implementation patterns discovered or established
• Insights about the codebase structure
• Blockers encountered and their resolutions
• Context that would be valuable for future sessions

Skip:

• Pleasantries and conversational filler
• Ephemeral states (e.g., "currently running tests")
• Information already stored in previous memories

For each significant item, call mcp__vector-memory__store_memories with appropriate metadata tags.

2. Set Waypoint

Call mcp__vector-memory__set_waypoint with:

project: [repository/project name]
branch: [current git branch]
summary: [2-3 sentences: what was the primary goal, what's the current status]

completed:
- [specific items with file paths where relevant]
- [e.g., "Implemented user auth in server/auth/login.ts"]

in_progress_blocked:
- [items still in flight with current state]
- [blockers with context on what's needed]

key_decisions:
- [decisions made and WHY - rationale is crucial]
- [e.g., "Chose JWT over sessions because X, Y, Z"]

next_steps:
- [concrete, actionable items]
- [prioritized if possible]

memory_ids:
- [IDs returned from store_memories calls above]

3. Report to User

Summarize what was stored:

• Number of memories created
• Key topics captured
• Waypoint stored confirmation

Guidelines

• Be thorough but concise - capture context that would take time to reconstruct
• Include file paths - makes resumption faster
• Explain decisions - future-you won't remember why
• Link memories - use memory_ids to connect related context
• Skip the obvious - don't store things easily discoverable from code