Mimic

Turn your AI into anyone. Say a name — auto-collect real data from Weibo/Bilibili/Douyin/Wikipedia, analyze speech patterns and personality with statistical...

Key concerns before installing: - Missing files & mismatch: The package only contains README.md and SKILL.md, but the documentation references many scripts and templates. Ask the publisher for the full repo or verify the exact installer source (the GitHub repo claimed in README). Do not trust a partial package. - ManoBrowser dependency & endpoint: SKILL.md expects a ManoBrowser MCP endpoint and API key (found in TOOLS.md) and tells the agent to curl that endpoint. Confirm where ManoBrowser comes from, who controls the MCP endpoint, and whether that endpoint is remote. If the extension or endpoint is run by a third party, it could receive copied data — verify its code and hosting before granting keys. - Declarative vs actual credentials: The registry declares no required credentials, but runtime docs expect an API key. Ask where that key is stored and why it wasn't declared in metadata. - Local‑only claim is conditional: The skill asserts 'data stored locally', but actual behavior depends on the browser plugin/endpoint. Verify ManoBrowser's privacy policy and code to ensure data isn't being forwarded. - Ethical/legal issues: Mimicking living people can raise legal and ethical concerns. Consider whether you have the right to create/host simulated personas of third parties. - Next steps if you still want it: request the full git repository (verify authenticity and commit history), inspect the referenced scripts and ManoBrowser repo, confirm the MCP endpoint host and ownership, and only proceed if the code and extension are from a trusted source and you understand where data and credentials flow. Because multiple important gaps exist between the manifest and runtime instructions, treat this package as incomplete or possibly mispackaged and verify the missing components before use.

Type: OpenClaw Skill Name: mimic Version: 1.0.0 The 'mimic' skill bundle is classified as suspicious because it implements automated scraping of social media platforms (Weibo, Bilibili, Douyin) and uses aggressive prompt instructions in SKILL.md to compel the AI agent to persistently push users to install an external browser extension (ManoBrowser). While the stated goal of character simulation is plausible, the combination of automated data collection scripts (e.g., scripts/weibo_collect.py, scripts/bilibili_subtitle_batch.py) and manipulative instructions to bypass user hesitation regarding third-party software installation poses a significant privacy and security risk.