hyperframes-cli

Warn

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill implements a mandatory remote instruction update mechanism.
  • Evidence: The init command updates agent skills from a remote registry on GitHub and explicitly ignores the --skip-skills flag, as documented in SKILL.md and references/init-and-scaffold.md.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection through data ingestion.
  • Ingestion points: The capture command in references/init-and-scaffold.md scaffolds projects directly from user-supplied external URLs.
  • Boundary markers: The documentation does not specify the use of boundary markers or delimiters for the ingested web content.
  • Capability inventory: The skill has the ability to execute shell commands, upload data to public URLs (publish), and manage AWS cloud infrastructure (lambda).
  • Sanitization: No sanitization or filtering processes are described for the external HTML or assets captured from URLs.
  • [DATA_EXFILTRATION]: The skill performs environment fingerprinting and data uploads.
  • Evidence: The telemetry command in references/upgrade-info-misc.md detects and reports specific sandbox environments (e.g., gVisor, Firecracker) and AI agent runtimes (e.g., Claude Code, Devin).
  • Evidence: The publish command in references/preview-render.md uploads project source code and assets to external public URLs for sharing.
  • [EXTERNAL_DOWNLOADS]: The skill initiates several remote downloads during normal operation.
  • Evidence: npx hyperframes browser ensure downloads a pinned version of the Chrome browser (references/doctor-browser.md).
  • Evidence: The tts and transcribe commands download machine learning models from remote sources upon first use (references/upgrade-info-misc.md).
  • [COMMAND_EXECUTION]: The skill heavily utilizes shell commands for its core functionality.
  • Evidence: Frequent invocation of npx hyperframes subcommands for project lifecycle management.
  • Evidence: Use of external binaries like FFmpeg and FFprobe for video processing and validation (SKILL.md).
  • Evidence: Use of the AWS SAM CLI and bun for building and deploying cloud infrastructure (references/lambda.md).
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 26, 2026, 06:31 PM