Manage WhatsApp via Evolution API: send messages, media, contacts, polls, manage groups, fetch chats, and summarize conversations.

This skill's behavior otherwise matches its description, but pay attention to these issues before installing: - Confirm the three environment variables (EVO_BASE_URL, EVO_INSTANCE_TOKEN, EVO_INSTANCE_NAME) will be provided by you; the skill will exit otherwise. The registry metadata omitted these — treat that as an authoring error and verify values before use. - Ensure EVO_BASE_URL points to a trusted Evolution API instance you control or trust. The script sends your EVO_INSTANCE_TOKEN in an HTTP header to that base URL — if the endpoint is malicious or public, the token could be misused. - Remove or sanitize any .env file in the skill directory (or modify the script) if you keep other secrets there; the script auto-sources .env which may load unrelated sensitive values. - Prefer using a token with minimal scope and rotate it if you test on shared infrastructure. If you need stronger assurance, review the Evolution API server code and host it yourself rather than using an unfamiliar hosted instance. If you want help: I can (a) produce a safer variant of evo.sh that refuses to source .env, (b) draft updated SKILL.md metadata to include the required env vars, or (c) checklist how to self-host and harden the Evolution API instance.

Type: OpenClaw Skill Name: evolution-whatsapp Version: 1.0.0 The skill is classified as suspicious due to critical shell injection vulnerabilities found in `evo.sh`. Multiple functions, such as `send_text`, `send_media`, and others, construct JSON payloads by directly embedding unsanitized user-supplied arguments into a string passed to `curl -d`. This allows for arbitrary command execution (RCE) on the host system if user input contains shell metacharacters. Additionally, the script attempts to load environment variables from a local `.env` file, which could be exploited if an attacker can control the file. While these are severe vulnerabilities, there is no clear evidence of intentional malicious behavior like data exfiltration to an attacker-controlled server or persistence mechanisms, aligning it with a 'suspicious' classification rather than 'malicious'.