Back to skill
Skillv1.0.0

ClawScan security

摸鱼日报和百度热搜等图片生成,持续升级中 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 11, 2026, 12:47 PM
Verdict
benign
Confidence
high
Model
gpt-5-mini
Summary
The skill's code, instructions, and requirements match its stated purpose (generating two poster types and exporting images); the main things to be aware of are outbound network requests to a third‑party API and remote image fetching, plus optional native image backends that are invoked if present on the system.
Guidance
This package appears to be what it says: a JSON-driven poster generator that renders SVG and optionally converts to PNG/JPG/WEBP. Before installing or running it, consider the following: - Network activity: the 'baidu_hot' renderer uses a third‑party API (https://v2.xxapi.cn/api/baiduhot) to fetch hot‑search data, and the daily renderer may fetch remote image URLs (and cache them under references/cache). If you need to avoid outbound requests, inspect and modify render_baidu_hot.py and render_daily_poster.py to use local data or a different API. - Image backends: conversion to PNG/JPG/WEBP may call system binaries (ImageMagick, Inkscape, librsvg 'rsvg-convert', resvg) or use the Python resvg_py/Pillow packages. Installing those packages/binaries grants the renderer the ability to execute local subprocesses for conversion — this is expected but worth noting. - Files written: the tool writes outputs to the path you pass and may create a cache directory (references/cache). Review the repo's references/ and cache if you are concerned about write locations. - Third‑party trust: the skill relies on an external API (XXApi) for baidu hot data; confirm you trust that service for the data you plan to publish. If privacy matters, replace or stub the API with local data. If you want a stricter safety posture, run the renderer in a restricted environment (no network, limited file write permissions) or review/patch the network fetch code to remove or control external calls.

Review Dimensions

Purpose & Capability
okName/description (generate 'daily' and 'baidu_hot' posters, output SVG and image exports) align with included files: rendering, lunar calendar, countdown, and SVG->image conversion. Declared requirements (none) are consistent with an instruction-first repo that includes Python scripts and a requirements.txt listing resvg_py and Pillow.
Instruction Scope
noteSKILL.md instructs the agent to run local Python renderer scripts and to produce SVG/PNG/JPG/WEBP outputs. The code legitimately reads JSON specs and writes output files. However, render_baidu_hot.py makes an outbound HTTP request to a third‑party endpoint (BAIDU_API_URL = "https://v2.xxapi.cn/api/baiduhot") to fetch hot‑search data, and render_daily_poster.py contains logic to fetch remote image assets (HTTP/HTTPS) and cache them under references/cache. SKILL.md does not explicitly call out network activity; if you need offline operation or want to avoid contacting external services, review or modify those places.
Install Mechanism
okNo automated install spec is provided (instruction-only), so nothing is downloaded or executed implicitly by an installer. The repository includes a requirements.txt (resvg_py, Pillow) which is proportional to SVG rendering and image conversion. Converters in svg_image_converter may invoke system binaries (ImageMagick 'magick', Inkscape, rsvg-convert, resvg) if present; this is expected behavior for image conversion but requires those binaries to be installed to enable some backends.
Credentials
okThe skill does not request environment variables, credentials, or configuration paths. There are no required secrets. Network calls are unauthenticated to a public API and to arbitrary image URLs supplied in spec; no credential exfiltration code is present.
Persistence & Privilege
okThe skill does not request permanent/enforced presence (always: false). It writes output files and may cache fetched images under references/cache, which is normal for this tool. It does not modify other skills or system-wide agent settings.