Reflexio

• Risk level: Low
• Analyzed: Apr 22, 2026, 7:42 AM
• Findings: 0
• Version: 1.0.10

Overview

The skill's files and runtime instructions are coherent with its stated purpose (local, Openclaw-based capture of user facts and playbooks); it asks for local config tweaks and uses LLM-driven prompts to read/merge/write .reflexio files but does not request unrelated credentials or remote downloads.

This plugin appears to do what it says: it extracts and stores local user facts and playbooks under .reflexio/ and uses Openclaw's memory/LLM capabilities to dedupe and consolidate. Before installing: (1) review the code and prompt templates if you don't trust their behavior, especially the LLM prompts that produce JSON used to write/delete files; (2) run it in a safe/test workspace first (backup or isolate any existing .reflexio/); (3) decline the automated config changes if you don't want the plugin to alter global active-memory or memorySearch.extraPaths (you can perform the steps manually after review); (4) do not paste secrets into conversations — the skill instructs redaction but relies on agent behavior to obey that rule; (5) if you require stricter controls, disable exec/automatic config changes and perform setup steps manually. Overall the package is internally coherent, but it has privileges to change Openclaw config and to write/delete local files — grant those only when you trust the plugin and have reviewed the prompts and code.