Personal Insight Engine

Personal Insight Engine (PIE) - A strategic analysis tool that scans local session logs (memory/*.md) and extracts 3 strategic insights using LLMs.

This skill appears to do what it says (scan memory/*.md, redact content, call an LLM). Before installing or running it: 1) Review your memory/ files — the scanner will read files under OPENCLAW_WORKSPACE or fall back to ./memory, so run it in an empty/safe workspace or move sensitive files out first. 2) Check your .env — load_dotenv will load any local .env file; remove or sanitise secrets you don't want processed. 3) Understand redaction is heuristic: inspect the clean_content implementation and test with non-sensitive sample files to confirm it removes the patterns you care about. 4) Note the Gemini API key is sent in the request URL (key=...) which is normal for Google API keys but could appear in logs; if that concerns you prefer the Zhipu provider. 5) If you need strict workspace guarantees, update the scanner to refuse the ./memory fallback and explicitly require OPENCLAW_WORKSPACE. If you want to proceed, run the code locally after review and avoid running it in an environment containing high-value secrets until you are confident in its redaction.

Type: OpenClaw Skill Name: personal-insight-engine Version: 1.0.4 The personal-insight-engine skill is a legitimate tool designed to analyze local session logs using LLMs (Gemini or Zhipu AI). It features a robust privacy-hardening layer in scripts/analyzer.py that uses regex to redact API keys, credentials, emails, IP addresses, and local file paths before data is transmitted. The code logic in scripts/pie.py and scripts/scanner.py is transparent, follows the stated purpose, and lacks any indicators of malicious intent or unauthorized data exfiltration.