LeoYeAI/agent-master-skills / skills/anima-aios
anima-aios
VerifiedAn AI Agent cognitive growth system built on the native OpenClaw architecture. It provides agents with persistent memory management, visual intimacy progression, a 5-dimensional cognitive profile, gamified daily quests, team leaderboards, and a 5-layer memory architecture with Knowledge Palace, Pyramid thinking, and Ebbinghaus decay function. 基于 OpenClaw 原生架构的 AI Agent 认知成长体系,为 Agent 提供五层记忆架构、知识宫殿、金字塔知识组织、记忆衰减函数、LLM 智能处理、永久化记忆管理、可视化亲密度成长、五维认知画像、游戏化每日任务和团队排行榜。
core/pyramid_engine.py
View on GitHub#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# Copyright 2026 Anima-AIOS Contributors
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Anima AIOS v6.0 - Pyramid Engine
金字塔知识组织引擎。实现自底向上的知识提炼:
实例层(instances)← 具体事件
→ 规则层(rules)← 从多个实例提炼的规律
→ 模式层(patterns)← 跨领域通用方法论
→ 本体层(ontology)← 核心概念和认知框架
移植自 Z 的 pyramid-abstract.js,适配 Python + LLM。
Author: 清禾
Date: 2026-03-23
Version: 6.0.0
"""
import os
import json
import hashlib
import logging
from pathlib import Path
from datetime import datetime
from typing import Dict, List, Optional, Literal
logger = logging.getLogger(__name__)
PyramidLevel = Literal["instances", "rules", "patterns", "ontology"]
class PyramidEntry:
"""金字塔条目"""
def __init__(self, entry_id: str, level: PyramidLevel, content: str,
topic: str = "", source_ids: List[str] = None,
created_at: str = "", metadata: Dict = None):
self.entry_id = entry_id
self.level = level
self.content = content
self.topic = topic
self.source_ids = source_ids or []
self.created_at = created_at or datetime.now().isoformat()
self.metadata = metadata or {}
def to_dict(self) -> Dict:
return {
"id": self.entry_id,
"level": self.level,
"content": self.content,
"topic": self.topic,
"source_ids": self.source_ids,
"created_at": self.created_at,
"metadata": self.metadata
}
class PyramidEngine:
"""
金字塔知识组织引擎
管理四层知识结构,支持自底向上的自动提炼。
"""
LEVELS = ["instances", "rules", "patterns", "ontology"]
LEVEL_UP = {"instances": "rules", "rules": "patterns", "patterns": "ontology"}
# 同一主题达到此数量时触发自动提炼
AUTO_DISTILL_THRESHOLD = 3
def __init__(self, agent_name: str, facts_base: str = None,
auto_distill: bool = False, llm_config: Dict = None):
self.agent_name = agent_name
self.pyramid_dir = Path(facts_base) / agent_name / "pyramid"
self.auto_distill = auto_distill
# LLM 客户端(可选)
self.llm = None
if llm_config:
try:
from distill_engine import LLMClient
self.llm = LLMClient(llm_config)
except ImportError:
pass
# 确保目录
self.pyramid_dir.mkdir(parents=True, exist_ok=True)
# 初始化各层文件
for level in self.LEVELS:
filepath = self.pyramid_dir / f"{level}.jsonl"
if not filepath.exists():
filepath.touch()
def _get_file(self, level: PyramidLevel) -> Path:
return self.pyramid_dir / f"{level}.jsonl"
def _read_entries(self, level: PyramidLevel) -> List[PyramidEntry]:
"""读取某一层的所有条目"""
filepath = self._get_file(level)
entries = []
if not filepath.exists():
return entries
for line in filepath.read_text(encoding='utf-8').splitlines():
line = line.strip()
if not line:
continue
try:
data = json.loads(line)
entries.append(PyramidEntry(
entry_id=data.get("id", ""),
level=data.get("level", level),
content=data.get("content", ""),
topic=data.get("topic", ""),
source_ids=data.get("source_ids", []),
created_at=data.get("created_at", ""),
metadata=data.get("metadata", {})
))
except json.JSONDecodeError:
continue
return entries
def _append_entry(self, entry: PyramidEntry):
"""追加写入一个条目"""
filepath = self._get_file(entry.level)
with open(filepath, 'a', encoding='utf-8') as f:
f.write(json.dumps(entry.to_dict(), ensure_ascii=False) + '\n')
def add_instance(self, content: str, topic: str, fact_id: str = "") -> PyramidEntry:
"""
添加实例(最底层)
Args:
content: 实例内容
topic: 主题分类
fact_id: 关联的 fact ID
"""
entry_id = f"inst_{datetime.now().strftime('%Y%m%d%H%M%S')}_{hashlib.md5(content.encode()).hexdigest()[:6]}"
entry = PyramidEntry(
entry_id=entry_id,
level="instances",
content=content,
topic=topic,
source_ids=[fact_id] if fact_id else []
)
self._append_entry(entry)
logger.debug(f"添加实例: {entry_id} (topic={topic})")
# 检查是否触发自动提炼
if self.auto_distill:
self._check_auto_distill(topic, "instances")
return entry
def _check_auto_distill(self, topic: str, current_level: PyramidLevel):
"""检查是否达到自动提炼阈值"""
entries = self._read_entries(current_level)
topic_entries = [e for e in entries if e.topic == topic]
if len(topic_entries) >= self.AUTO_DISTILL_THRESHOLD:
next_level = self.LEVEL_UP.get(current_level)
if next_level:
logger.info(f"主题 '{topic}' 在 {current_level} 层达到 {len(topic_entries)} 条,触发提炼到 {next_level}")
self.distill_up(topic, current_level)
def distill_up(self, topic: str, from_level: PyramidLevel) -> Optional[PyramidEntry]:
"""
向上提炼:从当前层提炼到上一层
Args:
topic: 主题
from_level: 源层级
Returns:
新创建的上层条目
"""
to_level = self.LEVEL_UP.get(from_level)
if not to_level:
logger.warning(f"已经是最高层: {from_level}")
return None
# 获取该主题在源层的所有条目
entries = self._read_entries(from_level)
topic_entries = [e for e in entries if e.topic == topic]
if not topic_entries:
return None
# 提炼内容
knowledge = self._distill_content(topic_entries, from_level, to_level)
if not knowledge:
return None
# 创建上层条目
entry_id = f"{to_level[:4]}_{datetime.now().strftime('%Y%m%d%H%M%S')}_{hashlib.md5(knowledge.encode()).hexdigest()[:6]}"
entry = PyramidEntry(
entry_id=entry_id,
level=to_level,
content=knowledge,
topic=topic,
source_ids=[e.entry_id for e in topic_entries]
)
self._append_entry(entry)
logger.info(f"提炼完成: {from_level} → {to_level} (topic={topic})")
return entry
def _distill_content(self, entries: List[PyramidEntry],
from_level: str, to_level: str) -> Optional[str]:
"""
提炼内容(优先 LLM,降级规则)
"""
combined = "\n".join([e.content[:200] for e in entries[:5]])
level_desc = {
"rules": "从多个具体案例中提炼出的规律和准则",
"patterns": "跨领域的通用方法论和思维模式",
"ontology": "最核心的概念定义和认知框架"
}
if self.llm:
prompt = f"""请从以下多条 {from_level} 级知识中,提炼出 {to_level} 级的知识。
{to_level} 的定义:{level_desc.get(to_level, '更高层次的抽象')}
源知识:
{combined}
请用简洁的语言总结,不超过 200 字。直接输出提炼结果。"""
response = self.llm.call(prompt, task="distill", max_tokens=300)
if response and len(response) > 10:
return response
# 规则降级:合并摘要
summaries = [e.content[:80] for e in entries[:3]]
return f"[{to_level}] 从 {len(entries)} 条 {from_level} 中提炼:\n" + "\n- ".join([""] + summaries)
def query(self, topic: Optional[str] = None, level: Optional[PyramidLevel] = None) -> List[Dict]:
"""
查询金字塔知识
Args:
topic: 按主题过滤
level: 按层级过滤
"""
results = []
levels = [level] if level else self.LEVELS
for lv in levels:
entries = self._read_entries(lv)
for entry in entries:
if topic and entry.topic != topic:
continue
results.append(entry.to_dict())
return results
def get_topics(self) -> Dict[str, Dict[str, int]]:
"""获取所有主题及其在各层的分布"""
topics = {}
for level in self.LEVELS:
entries = self._read_entries(level)
for entry in entries:
t = entry.topic or "_untagged"
if t not in topics:
topics[t] = {lv: 0 for lv in self.LEVELS}
topics[t][level] += 1
return topics
def get_stats(self) -> Dict:
"""获取金字塔统计"""
stats = {"agent": self.agent_name, "levels": {}, "total": 0}
for level in self.LEVELS:
count = len(self._read_entries(level))
stats["levels"][level] = count
stats["total"] += count
stats["topics"] = self.get_topics()
stats["auto_distill"] = self.auto_distill
return stats
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |