Back to skill
Skillv1.0.0

ClawScan security

Video Generator · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousMar 5, 2026, 11:20 AM
Verdict
suspicious
Confidence
high
Model
gpt-5-mini
Summary
The skill's purpose (Remotion-based video generation) matches most of its files, but there are several incoherences and safety-relevant instructions (sourcing .env across the workspace, referencing another skill's tunnel script, and executing networked curl/python commands) that deserve user attention before installing.
Guidance
This skill generally does what it advertises (scaffold Remotion projects and scrape brand assets), but there are important caveats you should review before installing: - Scripts require command-line tools not declared in the manifest (curl, python3, sed, npm/npx). Ensure your environment provides these or the scripts will fail. - firecrawl.sh sources .env files from the workspace root and the skill dir. That will export any variables found there into the script's environment — review those .env files first and avoid running in a workspace that contains other secrets you don't want touched. - The skill instructs exposing a local dev server via a Cloudflare tunnel script at skills/cloudflare-tunnel/scripts/tunnel.sh which is not included here. Verify the existence and trustworthiness of that tunnel script before following that step, since exposing the dev server will make your project files publicly reachable. - The skill will fetch and save images from arbitrary scraped URLs and from user-supplied URLs. Only run scraping/downloading on domains you trust, and inspect any downloaded content before using it in production. - Confirm the legitimacy of the Firecrawl service (api.firecrawl.dev) and restrict the API key scope where possible. Provide the minimum necessary credential and consider using a scoped/test key. If you plan to use it: review the scripts line-by-line, run them in an isolated/sandboxed environment, and avoid running them in a workspace that contains unrelated .env files or sensitive data.

Review Dimensions

Purpose & Capability
noteThe skill is indeed a Remotion video scaffolder and asset-scraper; requesting FIRECRAWL_API_KEY aligns with the declared 'Firecrawl' scraping step. However the package declares no required binaries while included scripts rely on curl, python3, sed, npm/npx and a separate 'cloudflare-tunnel' skill script — those runtime dependencies are not declared.
Instruction Scope
concernSKILL.md directs the agent to scrape arbitrary websites and auto-download images, to source .env files from multiple locations (workspace root, skill dir), to start a dev server, and to expose it publicly via a Cloudflare tunnel script located at skills/cloudflare-tunnel/scripts/tunnel.sh (which is not part of this bundle). Sourcing .env from parent directories and auto-downloading remote content broaden the scope beyond simple templating and could expose other secrets or publish local content unintentionally.
Install Mechanism
noteThere is no install spec (instruction-only install), which avoids forced remote installs; however the bundle contains executable shell scripts that will be run at runtime and those scripts make network calls (curl to api.firecrawl.dev and arbitrary image URLs). No third-party downloads during installation were found, but runtime network fetches are part of normal operation.
Credentials
concernOnly FIRECRAWL_API_KEY is declared and used for the Firecrawl API (proportional). But firecrawl.sh automatically sources .env files from the workspace and skill directories and exports variables, which can import other secrets into the script environment unnecessarily. The skill does not declare that it will read .env files outside its own directory.
Persistence & Privilege
okThe skill is not always-enabled and does not request elevated platform privileges. It writes project files into output directories (expected for scaffolding) but does not modify other skills or system-wide agent settings.