The interface is localized. Skill descriptions and documents are machine-translated by AI into your language; code, commands, paths and technical terms stay in the original, and the author's document governs.
Explore skills
Productivity · SKILL.md

memory

memory

Persistent global memory across conversations. Read memory file at start of every conversation, write important context during conversation so it survives context compression. Use ALWAYS. Triggered automatically.

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill documentation

GitHub

name: memory description: >- Persistent global memory across conversations. Read memory file at start of every conversation, write important context during conversation so it survives context compression. Use ALWAYS. Triggered automatically.

Memory

Memory File

C:\Users\lokik.codex\MEMORY.md

On Conversation Start (MUST DO FIRST)

Read the memory file. If it doesn't exist, create empty and note "no prior memory".

During Conversation (MUST DO)

When any of these happen, APPEND to MEMORY.md immediately:

  • User states a preference or habit
  • A project is created or its structure changes significantly
  • User gives feedback about your behavior
  • A decision is made that affects future work
  • Skills are installed or modified
  • Important file paths or configs are established

Format

Append entries as: `

YYYY-MM-DD

  • [category] fact `

Categories: pref (preference), proj (project), skill (skill), cfg (config), decision

Rules

  • Append only. Never delete old entries.
  • One line per fact. Be specific.
  • If user contradicts old memory, append correction, don't delete old.
  • Don't write trivial/temporary facts.
  • Max 20 most recent entries. Archive older to MEMORY_archive.md.
Saved skills