The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform various shell-based operations using the hyperframes CLI via npx, including project linting, validation, asset rendering, and text-to-speech synthesis.
[COMMAND_EXECUTION]: The documentation provides a 'ladder of degradation' for task execution, which includes instructions to start CLI children as background shell processes and redirect their input/output if native agent delegation tools are unavailable.
[INDIRECT_PROMPT_INJECTION]: The skill implements an advanced sub-agent dispatch system where the agent is instructed to ingest 'role files' and 'dispatch context' verbatim to spawn child agents. This architectural pattern creates a surface for potential indirect prompt injection if the project's source files are modified by an untrusted party.
Ingestion points: agents/<role>.md files and '## Dispatch context' blocks within project workflows.
Boundary markers: The instructions specify placing ingested data into a designated '## Dispatch context' section to separate it from the system prompt.
Capability inventory: Subprocess spawning (CLI workers), file system write access (writing child prompts to disk), and environment-specific delegation tools (Agent, spawn_agent, delegate_task).
Sanitization: No sanitization or escaping is mentioned for the ingested data; the skill emphasizes copying content verbatim.
[EXTERNAL_DOWNLOADS]: The project templates reference the GSAP animation library from cdn.jsdelivr.net and the Tailwind CSS browser runtime from well-known technology providers.

hyperframes-core