honest-forget

name	honest-forget
description	Graceful memory compression with integrity — summarize before forgetting, never fabricate
license	MIT
tier	1
allowed-tools	["read_file","write_file","list_dir","search_replace"]
related	["robust-first","summarize","self-repair","session-log","memory-palace","postel"]
tags	["moollm","memory","compression","graceful","context"]
inputs	{"target":{"type":"path","required":true,"description":"What to potentially forget"},"urgency":{"type":"enum","values":["low","medium","high","critical"],"default":"medium","description":"How urgently we need to free context"}}
outputs	["FORGET.yml","WISDOM.yml","POINTERS.yml","archive/"]
templates	["FORGET.yml.tmpl","WISDOM.yml.tmpl"]

🌫️ Honest-Forget Skill

"Summarize before forgetting. Never fabricate."

Graceful memory compression that preserves wisdom. Compress context gracefully when budget is exceeded. Extract wisdom, create pointers, and let go with integrity.

Purpose

Compress context gracefully when budget is exceeded. Extract wisdom, create pointers, and let go with integrity. Never silently lose information or fabricate what was forgotten.

When to Use

• Context window is filling up
• Completed work needs archiving
• Repetitive iterations need compression
• Old sessions need summarization
• "I've tried this 10 times" situations

The Honest Forget Cycle

ASSESS → EXTRACT → COMPRESS → POINTER → RELEASE

Protocol

Assessment

Before forgetting, understand what you have:

assessment:
  file: "path/to/file.md"
  tokens: 5000
  
  contains:
    decisions: ["List of decisions made"]
    learnings: ["What was learned"]
    questions_answered: ["Q&A pairs"]
    dead_ends: ["What didn't work"]
    
  importance:
    for_current_task: "high|medium|low"
    for_future_reference: "high|medium|low"

Wisdom Extraction

Compress iterations into lessons:

wisdom:
  id: "wisdom-001"
  title: "LEARNED: [Pattern/Pitfall]"
  
  compressed_from:
    iterations: "45-55"
    original_tokens: 15000
    
  lesson: