VeriTeknik

memory-workflow

"Guide the memory session lifecycle - start sessions, record observations, search memories, and end sessions with Z-reports. Use when working with Plugged.in memory system."

VeriTeknik 7 1 Updated 2mo ago
GitHub

Install

npx skillscat add veriteknik/pluggedin-plugin/memory-workflow

Install via the SkillsCat registry.

SKILL.md

Memory Workflow

Manage the Plugged.in memory session lifecycle.

Session Lifecycle

  1. Start Session: Call pluggedin_memory_session_start at the beginning of work
  2. Observe: During work, call pluggedin_memory_observe to record:
    • tool_call / tool_result: Tool usage and outcomes
    • user_preference: Explicit user preferences
    • error_pattern: Errors and their resolutions
    • success_pattern: What worked well
    • decision: Key decisions made
    • insight: Conclusions worth remembering
  3. Search: Use pluggedin_memory_search to find relevant past memories (returns lightweight summaries)
  4. Details: Use pluggedin_memory_details for full content of specific memories
  5. End Session: Call pluggedin_memory_session_end to generate a Z-report

Arguments

  • $ARGUMENTS = "start": Start a new memory session
  • $ARGUMENTS = "end": End the current session with Z-report
  • $ARGUMENTS = "status": Show current session status

What to Observe

Record observations for anything that would be useful in future sessions:

  • Error patterns and their solutions
  • User preferences about workflow, tools, or coding style
  • Successful approaches to recurring problems
  • Key architectural decisions and their rationale
  • Tool configurations that worked

Progressive Disclosure

Memory search uses a 3-layer system for token efficiency:

  1. Search (Layer 1): Returns summaries (50-150 tokens each)
  2. Timeline (Layer 2): Adds temporal context
  3. Details (Layer 3): Returns full content

Always start with search. Only request details for memories you actually need.