Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
clawbot-drama-producer
v1.0.2皮皮虾职场短剧全流程制作技能。用于为「皮皮虾」（机械龙虾AI-bot）职场短剧生成镜头视频、剪辑成片、配音配乐并发布到飞书群。完整流程：图生视频(I2V) → ffmpeg规范化+剪辑 → TTS配音 → BGM混音 → 飞书媒体消息发送。当用户提到制作皮皮虾短剧、生成新镜头、剪辑视频、配音配乐、或将视频/音频发...
⭐ 0· 282·0 current·0 all-time
byRong@kylinr
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The name/description (generate I2V, normalize/trim, TTS, mix audio, post to Feishu) aligns with the included scripts and instructions. However the SKILL.md and scripts assume the presence of specific binaries/paths and an agent config file (e.g., /workspace/bin/ffmpeg, /app/openclaw/node_modules/.bin/node-edge-tts, /root/.openclaw/openclaw.json) but the skill declares no required env vars or config paths — a mismatch between claimed requirements and actual runtime needs.
Instruction Scope
Runtime instructions and scripts explicitly read /root/.openclaw/openclaw.json to extract Feishu appId/appSecret, generate tenant tokens, upload media, and send messages. They also read/write files under /workspace and /tmp and call external endpoints (Feishu API and incompetech for BGM). Reading agent config and tenant credentials is outside a minimal 'video production' description unless sending to Feishu is explicitly declared with required credential access.
Install Mechanism
No install spec (instruction-only with helper scripts). That minimizes installation risk because nothing is fetched or executed at install time, but the runtime scripts will execute local binaries and network calls.
Credentials
The skill does not declare any required environment variables or config paths, yet the scripts extract appId/appSecret from /root/.openclaw/openclaw.json and call Feishu APIs. Accessing tenant credentials and agent configuration is a high-privilege operation relative to a content-production task and should be explicitly declared and consented to. The skill also expects ffmpeg/ffprobe and node-edge-tts at specific paths (implicit requirements).
Persistence & Privilege
always:false and no self-enabling behavior is present, which is good. However the skill reads the agent's global config (/root/.openclaw/openclaw.json) to obtain credentials for Feishu; reading another component's config expands its effective privilege and blast radius even though it does not persist or alter other skills. This is a notable privilege surface to surface to the user.
Scan Findings in Context
[reads-openclaw-config] expected: The send_audio.sh and send_video.sh scripts load /root/.openclaw/openclaw.json to extract feishu appId/appSecret. Reading credentials is necessary to post to Feishu but the skill did not declare config paths or required credentials.
[calls-feishu-api] expected: Scripts perform tenant token request and upload/send media to https://open.feishu.cn/open-apis/* — this is consistent with the stated goal of publishing to Feishu, but it will transmit media and text to an external group using tenant credentials.
[uses-local-ffmpeg-and-edge-tts] expected: The skill expects ffmpeg/ffprobe at /workspace/bin and node-edge-tts at /app/openclaw/node_modules/.bin/node-edge-tts. These binaries are required for video/audio work but were not declared in the skill metadata.
What to consider before installing
This skill appears to implement the video/TTS/mix/send workflow described, but it silently reads your agent's global config (/root/.openclaw/openclaw.json) to obtain Feishu appId/appSecret and then uses those credentials to upload media and post messages. Before installing or enabling it: (1) Inspect /root/.openclaw/openclaw.json to confirm what credentials and scopes would be used and whether you want this skill to access them; (2) Prefer creating a dedicated Feishu bot/account with minimal permissions for publishing content instead of reusing tenant/global credentials; (3) Ensure ffmpeg, ffprobe and node-edge-tts are present at the expected paths or update the scripts to point to approved binaries; (4) Consider running the skill in a sandboxed workspace so it cannot read other sensitive files; (5) If you want stricter controls, ask the author to declare required config paths/env vars (or refactor to accept explicit credentials via secure input) so consent is explicit. If you want me to, I can suggest concrete changes to the scripts to avoid reading global config and to document required credentials.Like a lobster shell, security has layers — review code before you run it.
dramavk9778qww4zfk0ez6asaqa6sbn582gdfnfeishuvk9778qww4zfk0ez6asaqa6sbn582gdfnlatestvk9778qww4zfk0ez6asaqa6sbn582gdfnopenclawbotvk9778qww4zfk0ez6asaqa6sbn582gdfnvideovk9778qww4zfk0ez6asaqa6sbn582gdfn
License
MIT-0
Free to use, modify, and redistribute. No attribution required.