Agent Skills
› omega-memory/omega-memory
› omega-memory
omega-memory
GitHub为AI编程代理提供持久化记忆层,支持跨会话决策继承、错误经验学习及任务断点续传。通过本地语义搜索和上下文重排序,帮助代理记住上下文并避免重复犯错。
Trigger Scenarios
需要跨会话保持上下文连续性
希望代理从历史错误或决策中学习
需要暂停并稍后恢复复杂编码任务
Install
npx skills add omega-memory/omega-memory --skill omega-memory -g -y
SKILL.md
Frontmatter
{
"name": "omega-memory",
"version": "1.0.0",
"description": "Persistent memory for AI coding agents. Semantic search, auto-capture, checkpoint\/resume across sessions.",
"requires_env": [],
"requires_binaries": [
"python3",
"pip3"
]
}
OMEGA Memory
Persistent memory for AI coding agents. Your agent remembers decisions, learns from mistakes, and picks up where it left off.
Installation
pip3 install omega-memory[server]
omega setup
The omega setup command auto-configures your MCP client (Claude Code, Cursor, Windsurf, or Zed). No API keys needed — runs fully local with CPU-only embeddings.
What It Does
OMEGA gives your agent a persistent memory layer across coding sessions:
- Decisions & context carry forward — no re-explaining
- Lessons learned from errors are recalled before you repeat them
- Checkpoint/resume lets you pause complex tasks and pick up later
- Semantic search over all stored memories with contextual re-ranking
MCP Tools (12 tools)
| Tool | Purpose |
|---|---|
omega_welcome |
Session briefing with recent memories and profile |
omega_protocol |
Retrieve operating rules and behavioral guidelines |
omega_store |
Store typed memory (decision, lesson, error, preference, summary) |
omega_query |
Semantic or phrase search with tag filters and re-ranking |
omega_lessons |
Cross-session lessons ranked by access count |
omega_profile |
Read or update the user profile |
omega_checkpoint |
Save task state for cross-session continuity |
omega_resume_task |
Resume a previously checkpointed task |
omega_memory |
Manage a specific memory (edit, delete, feedback, similar, traverse) |
omega_remind |
Set, list, or dismiss time-based reminders |
omega_maintain |
System housekeeping (health, consolidate, compact, backup, restore) |
omega_stats |
Analytics: type breakdown, session stats, weekly digest, access rates |
Usage Pattern
At the start of every session:
- Call
omega_welcome()for context briefing - Call
omega_protocol()for operating instructions - Follow the protocol it returns
During work:
- Before non-trivial tasks:
omega_query()to check for prior context - After completing tasks:
omega_store(content, "decision")to save key outcomes - When context is getting full:
omega_checkpoint()to save state
Links
Version History
- 50dffeb Current 2026-07-24 16:27


