The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch and update components using npx hyperframes@latest and npx skills add heygen-com/hyperframes. These target the official HyperFrames NPM package and the HeyGen GitHub repository.
[REMOTE_CODE_EXECUTION]: The skill facilitates remote code execution by invoking the HyperFrames CLI via npx and running workflow-specific scripts (Python and Node.js) that are installed lazily during the project lifecycle.
[COMMAND_EXECUTION]: The agent is directed to execute various shell commands, including npx for CLI tools, gh for fetching GitHub Pull Request metadata, and several internal scripts like analyze-beatgrid.py and transcribe.mjs for media processing.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection as it ingests and processes untrusted data from external sources.
Ingestion points: Data is sourced from website URLs (via headless capture), GitHub PR references (via gh pr view), and Figma design files (via REST/CLI import).
Boundary markers: The instructions do not specify the use of clear delimiters or instructions for the agent to ignore potentially malicious prompts embedded within the external content being processed.
Capability inventory: The skill environment allows for shell command execution, network access (via CLIs), and file system modifications, which could be leveraged if an injection is successful.
Sanitization: While the skill mentions that Figma imports are sanitized, there is no explicit instruction for the sanitization of content retrieved from arbitrary URLs or GitHub PR descriptions.

hyperframes