pancake-skills

PassAudited by VirusTotal on May 12, 2026.

Overview

Type: OpenClaw Skill Name: pancake-skills Version: 1.0.0 The skill provides comprehensive functionality to interact with the Pancake Platform API, including read and write operations for pages, conversations, messages, customers, and media uploads. While these capabilities are powerful, the skill includes explicit safety features such as the `CONFIRM_WRITE=YES` guardrail for all write operations and clear instructions in `SKILL.md` and `README.md` that guide the AI agent towards safe usage and warn against common pitfalls (e.g., not saving tokens). The code is straightforward shell scripting without any malicious constructs, obfuscation, or attempts at prompt injection to subvert the agent's behavior for harmful purposes. All API calls are directed to the legitimate `https://pages.fm` domain.

Findings (0)

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

If a user token is available, the agent could create or refresh a page access token without the user setting CONFIRM_WRITE=YES for that action.

Why it was flagged

This POST operation generates or refreshes a page access token but does not call confirm_write, unlike other visible write commands. That bypasses the documented write-safety model for a high-impact credential action.

Skill content
pages-generate-token)
    require_env USER_ACCESS_TOKEN
    page_id="${1:?PAGE_ID required}"
    pancake_request_user POST "/api/v1/pages/${page_id}/generate_page_access_token"
    ;;
Recommendation

Add confirm_write before pages-generate-token, and treat token generation as a privileged write operation requiring explicit user approval.

What this means

Anyone or any agent process with these tokens can access and modify Pancake page/customer/message data within the token's permissions.

Why it was flagged

The skill needs personal and page-level Pancake tokens, including a non-expiring page token. This is purpose-aligned, but it is broad account authority and the registry metadata declares no primary credential or required env vars.

Skill content
`USER_ACCESS_TOKEN`: Token cá nhân... Có hiệu lực 90 ngày.

`PAGE_ACCESS_TOKEN`: Token của page... Không hết hạn trừ khi xóa/renew.
Recommendation

Use the least-privileged Pancake tokens available, avoid sharing them broadly, rotate page tokens when no longer needed, and update metadata to declare the credential requirements.

What this means

Unusual or attacker-influenced token/date values could cause local code execution when the script URL-encodes them.

Why it was flagged

The helper interpolates an input string directly into Python source code. A crafted value containing a quote could break out of the string and execute unintended Python code under the user's account.

Skill content
url_encode() {
  python3 -c "import urllib.parse; print(urllib.parse.quote('$1', safe=''))"
}
Recommendation

Rewrite url_encode to pass data as an argument or stdin, for example `python3 -c 'import sys, urllib.parse; print(urllib.parse.quote(sys.argv[1], safe=""))' "$1"`.