PIE (Personal Insight Engine)
Analyze your startup journey by distilling strategic patterns from your memory logs.
Setup
1. Configure API Keys:
PIE requires at least one of the following environment variables. If both are provided, Gemini is used by default.
- -
GEMINI_API_KEY: For Google Gemini 2.0 Flash (Recommended). - INLINECODE1 : For Zhipu AI GLM-4.
2. Dependencies:
CODEBLOCK0
Usage
Run Weekly Review
By default, scans the last 7 days and auto-detects the provider:
CODEBLOCK1
Force a Provider
If you want to use a specific LLM regardless of your environment defaults:
CODEBLOCK2
Custom Lookback
Scan a specific number of days (e.g., last 30 days):
CODEBLOCK3
Privacy & Safety (Hardened)
PIE is built for high-stakes startup analysis where privacy is non-negotiable:
- - Zero-Exposure Redaction: Before transmission, all content is scrubbed for:
- API Keys, Tokens, and Passwords (regex-hardened).
- Email addresses and IP addresses.
- Local file system paths (home/user directories).
- - Standardized Pathing: Only scans the
memory/ folder within the official OPENCLAW_WORKSPACE. - Content Distillation: Strips out JSON metadata and system headers to ensure the LLM only sees the semantic dialogue.
How it works
- 1. Discovery: Standardized scan of
memory/ via OPENCLAW_WORKSPACE. - Sanitization: Multi-stage PII redaction (IPs, Emails, Keys, Paths).
- Synthesis: Pattern recognition across logs for strategic pivots and pain points.
- Output: Formatted Markdown report for the startup founder.
Developed by Leo & Neo (Startup Partners)
PIE(个人洞察引擎)
通过从你的记忆日志中提炼战略模式,分析你的创业历程。
设置
1. 配置 API 密钥:
PIE 至少需要以下一个环境变量。如果同时提供两者,默认使用 Gemini。
- - GEMINIAPIKEY:用于 Google Gemini 2.0 Flash(推荐)。
- ZHIPUAPIKEY:用于智谱 AI GLM-4。
2. 依赖项:
bash
pip install -r requirements.txt
使用方法
运行每周回顾
默认扫描最近 7 天并自动检测提供商:
bash
python3 scripts/pie.py
强制指定提供商
如果你想忽略环境默认值,使用特定的 LLM:
bash
python3 scripts/pie.py --provider zhipu
python3 scripts/pie.py --provider gemini
自定义回溯天数
扫描特定天数(例如,最近 30 天):
bash
python3 scripts/pie.py --days 30
隐私与安全(强化版)
PIE 专为隐私不可妥协的高风险创业分析而构建:
- API 密钥、令牌和密码(正则表达式强化)。
- 电子邮件地址和 IP 地址。
- 本地文件系统路径(主目录/用户目录)。
- - 标准化路径:仅扫描官方 OPENCLAW_WORKSPACE 内的 memory/ 文件夹。
- 内容提炼:剥离 JSON 元数据和系统标头,确保 LLM 仅看到语义对话。
工作原理
- 1. 发现:通过 OPENCLAW_WORKSPACE 对 memory/ 进行标准化扫描。
- 净化:多阶段 PII 编辑(IP、电子邮件、密钥、路径)。
- 综合:跨日志进行模式识别,寻找战略转折点和痛点。
- 输出:为创业创始人生成格式化的 Markdown 报告。
由 Leo 和 Neo(创业合伙人)开发