memori-extension

Memory augmentation and LLM call interception using the Memori Python library with optional Zhipu API integration.

This skill appears coherent, but follow these precautions before enabling external features: 1) Test in local-only mode by leaving ZHIPUAI_API_KEY unset—this keeps all work on your machine. 2) Review the Memori and zhipuai packages (origin, maintainers) before pip installing. 3) If you do supply ZHIPUAI_API_KEY, understand that full conversation text (including system prompts and assistant responses) may be transmitted to Zhipu's servers. 4) Protect the local database and tech_terms file with appropriate file permissions and backups; the skill can append terms to the configured file. 5) If you need higher assurance, run the skill in a sandbox environment and inspect traffic to confirm only intended API calls occur.

Type: OpenClaw Skill Name: memori-extension Version: 1.0.2 The skill is transparent about its operations, explicitly detailing local file read/write (SQLite database, tech terms file) and optional external API calls to Zhipu AI if an API key is provided. The `SKILL.md` and `README.md` provide clear security and privacy notices, warning users about data transmission if the Zhipu AI API is enabled. The Python code (`memori_extension.py`) implements these features as described, only making external calls when `ZHIPUAI_API_KEY` is set. There is no evidence of intentional malicious behavior such as unauthorized data exfiltration, credential theft, backdoors, or prompt injection designed to subvert the agent's purpose. All risky capabilities are clearly documented and opt-in.