Search

PythonReactSEOPDF
GitHub Skill

memory-keeper

Manages project structure, auto-saves artifacts, and performs recovery protocols. Use this skill at the end of tasks or when encountering errors.

mr-phariyawit☆ 0 Repository stars0 Repository forks

Installation

Follow the repository instructions for this skill and your agent. Review the files and dependencies before adding it to your workflow.

View source ↗Guides

Skill instructions

Memory Keeper Skill

Purpose

To ensure the Structural Memory of the agent is preserved, artifacts are backed up, and the agent can recover from context limits.

Trigger Conditions

  • Task completion (finishing a feature).
  • User mentions /retro or "save memory".
  • Agent encounters "Agent terminated due to error" or "Context Limit" errors.

Workflow Rules

1. Structural Integrity Check
  • Verify agent/rules/ exists.
  • Verify docs/ exists.
  • Verify .memory/ exists.
    • Action: If missing, suggest running init-project or creating them.
2. Auto-Save Protocol

When a major task is done:

  1. Format Timestamp: YYMMDD_HHMM.
  2. Target Directory: .memory/[Timestamp]_[TaskName]/.
  3. Action: Copy key artifacts (task.md, implementation_plan.md, spec.md) to this folder.
    • Note: Do not perform the copy yourself if it's too heavy. Just remind the user or create a script.
3. Recovery Protocol (Agent Terminated Handler)

If the Agent is slow, buggy, or hits token limits:

  1. Advise User: "I am hitting context limits. Recommendation: Factory Reset."
  2. Command: Suggest running:
    ./antigravity_toolkit.sh full
    
  3. Post-Reset: Remind user to:
    • "Import rules and workflows" immediately after reset.

Examples

Read the usage examples in the source repository.

View source ↗

Source history

Indexed on
2026-01-20
Index updated
2026-01-26
Branch
main

Index and sync dates describe this directory. View the source repository for code history.

GitHub ↗

chat Comments (0)

chat_bubble_outline

No comments yet. Be the first to share your thoughts!