hyperframes

{"verdict":"SAFE","summary":"The 'hyperframes' skill is a comprehensive documentation and instruction set for a legitimate video rendering framework. The skill teaches an AI agent how to author, edit, and render videos using HTML, CSS, and JavaScript. The analysis confirms that all external resources and instructions are consistent with the documented purpose and originate from a legitimate vendor. No malicious patterns were detected.","categories":["SAFE"],"detailed_analysis":"- [PROMPT_INJECTION]: No evidence of malicious prompt injection was found. The instructions contain natural organizational markers, but these are used to guide the agent's logic for choosing workflows and ensuring technical correctness. There are no attempts to bypass safety filters or ignore previous instructions.

• [DATA_EXFILTRATION]: No malicious exfiltration patterns were detected. While the documentation mentions sensitive paths like ~/.aws/credentials, it is in the context of user prerequisites for AWS Lambda deployment. The skill does not attempt to read these files. Network operations target the vendor's own GitHub repository for legitimate purposes.
• [REMOTE_CODE_EXECUTION]: The skill uses npx to run its own CLI tools. These are standard operational commands for the framework. External scripts are loaded via well-known and trusted CDNs (e.g., jsdelivr.net for GSAP and Three.js). The lambda workflow involves deploying code to the user's own AWS account, which is an intended and documented feature.
• [COMMAND_EXECUTION]: The skill instructs the agent to use standard CLI tools (npx, git, gh, curl) necessary for its operation. These commands are used within the scope of the skill's primary purpose.
• [OBFUSCATION]: No obfuscated content, Base64-encoded commands, zero-width characters, or homoglyphs were found in the provided files.
• [DYNAMIC_EXECUTION]: The skill's primary function involves authoring HTML/JS code for video rendering. While this involves dynamic execution in a browser or headless Chrome environment, it is the core, documented purpose of the framework. No unsafe deserialization or unauthorized code injection patterns were identified.
• [PRIVILEGE_ESCALATION]: No attempts to gain elevated privileges were found. The use of sudo or similar commands is absent.
• [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface as it ingests external data (website content, GitHub PRs, user notes) to generate video compositions. However, the documentation includes multiple validation steps that mitigate accidental errors. The risk is assessed as low due to these technical checks and the nature of the output.
• [CREDENTIALS_UNSAFE]: No hardcoded credentials or unsafe secret handling instructions were found. The skill correctly points towards standard AWS credential management.","python_packages":[],"node_packages":["hyperframes","@hyperframes/aws-lambda","@hyperframes/core","gsap","three","tailwindcss"],"remote_code_patterns":["https://raw.githubusercontent.com/