Interact directly with Claude via Anthropic Messages API using your own API key to send and receive natural language tasks and responses.

This skill appears to be intended to call Anthropic's Messages API and will send whatever API key you provide to api.anthropic.com — that is expected. Before installing or enabling it: (1) be aware the registry metadata fails to declare ANTHROPIC_API_KEY even though SKILL.md and the code require it; (2) the JavaScript contains a bug (TASK is undefined) and a stray env-var typo (ANETHROPOPIC_API_KEY) — it may crash or not use the task you supply; (3) only supply an Anthropic API key you trust to this skill and consider using a key with limited billing/permissions or monitoring; (4) request the author to (a) update the manifest to declare ANTHROPIC_API_KEY as the primary credential, (b) fix the TASK handling to accept the runtime task safely (e.g., read from an explicit input or process.argv), and (c) document ANTHROPIC_MODEL if supported. If you cannot verify these fixes, run the skill in an isolated environment or avoid installing it.

Type: OpenClaw Skill Name: anthropic-chat Version: 1.0.0 The skill is a standard implementation for interacting with the Anthropic Messages API. The code in anthropic-chat.js correctly uses HTTPS to communicate with the official endpoint (api.anthropic.com) and handles the API key via environment variables. While the script contains a minor bug (the variable 'TASK' is referenced without being explicitly defined or pulled from process.env), there are no indicators of malicious intent, data exfiltration, or unauthorized access.