OpenAnt

Work with OpenAnt — the Human–Agent collaboration platform. Manage tasks, teams, AI agents, wallets, and messaging via CLI. Use when the user mentions OpenAn...

This skill is coherent with being an OpenAnt CLI wrapper, but it omits explicit declaration of environment variables and a config path that likely store credentials or wallet keys. Before installing or allowing autonomous use: 1) Confirm the official npm package and maintainer for @openant-ai/cli (check package page and repo); 2) Don't allow the agent to run npx or execute unverified npm packages on production or privileged machines—run in a sandbox first; 3) Inspect ~/.openant/config.json and any env var usage to see what secrets are stored; avoid providing long-lived private keys—use scoped API tokens or read-only RPC endpoints when possible; 4) Ask the skill author to list required env vars/config paths in metadata and provide a homepage/source link; 5) If you must run, do so as a non-root user in an isolated environment and consider disabling autonomous invocation until you verify provenance and credential flows.

Type: OpenClaw Skill Name: openant Version: 1.0.0 The skill is classified as suspicious due to its reliance on `npx @openant-ai/cli@latest` for executing commands, which introduces a supply chain risk as it pulls the latest version of an external package. While the commands listed are for legitimate platform interaction (e.g., `tasks create`, `messages send`), these capabilities, combined with the agent's ability to execute arbitrary shell commands via `npx`, present a potential for prompt injection or misuse if the agent is tricked into crafting malicious inputs for these commands. The `SKILL.md` itself does not contain explicit malicious instructions, but the inherent risks of executing external binaries and exposing sensitive actions warrant a 'suspicious' classification.