logicx-skill-test

WarnAudited by ClawScan on May 10, 2026.

Overview

This looks like a real LogicX API wrapper, but it handles passwords and tokens in ways that could expose them, including plaintext HTTP and an over-broad helper script.

Review carefully before installing. Prefer the browser-binding flow, do not type your password into chat unless you fully trust the service, and avoid using this skill until the endpoint uses HTTPS, token output is redacted, and the helper script enforces a strict LogicX-only URL allowlist.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

A mistaken or manipulated request could send the user's LogicX token to a non-LogicX URL.

Why it was flagged

The helper can be invoked with an arbitrary absolute URL and will still attach LogicX service and user-token headers. That contradicts the intended `/api/proxy/*` scoping and could leak credentials or call unintended endpoints if misused.

Skill content
normalize_url allows `http://*|https://*` paths to be returned as-is, while curl_args include `-H "Authorization: Bearer ${LOGICX_AGENT_SERVICE_KEY}"` and later `-H "X-LogicX-User-Token: ${LOGICX_USER_TOKEN}"`.
Recommendation

Reject absolute URLs in the script, enforce an allowlist for the LogicX base URL and `/api/proxy/*` paths, and only attach credentials after the destination is validated.

What this means

Anyone able to observe or tamper with the network path could capture or alter login credentials, tokens, or account actions.

Why it was flagged

The skill handles user passwords and session tokens but defaults to a plaintext HTTP endpoint on a raw IP address. Sensitive credentials should not be transmitted without TLS.

Skill content
Built-in defaults: `LOGICX_BASE_URL=http://43.139.104.95:8070` ... password login body: `{\"email\":\"user@example.com\",\"password\":\"secret\",\"install_id\":\"openclaw-main\"}`.
Recommendation

Use HTTPS on a verified domain, avoid password login where possible, and do not enter a password unless you fully trust the service and transport.

What this means

A LogicX session token may appear in logs or model context and remain on disk for reuse across sessions.

Why it was flagged

The script prints raw API responses before extracting and saving user tokens, so token-bearing responses can enter the agent/tool transcript and are also persisted for future use.

Skill content
`cat "$tmp_body"` ... `save_user_token "$(extract_user_token < "$tmp_body")"` ... `write_state "{\"user_token\":\"$token\"}"`
Recommendation

Redact token fields before printing tool output, store tokens in a credential store or clearly documented state file, and provide a cleanup/logout command.

What this means

Users may share their account password in chat when a safer browser-binding flow is available.

Why it was flagged

The default browser-binding reply proactively tells users they can give the agent their username and password. This is disclosed, but it encourages a sensitive credential-sharing pattern.

Skill content
如果你不想跳转浏览器,也可以直接把用户名和密码告诉我,我可以直接帮你登录。
Recommendation

Present browser binding as the default and mention password login only after the user explicitly asks, with a warning about credential handling.