The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run sudo 5dive agent import to provision new agent instances, which executes commands with elevated system privileges.
[EXTERNAL_DOWNLOADS]: The skill references and downloads assets from GitHub repositories (5dive-ai/openagent, 5dive-ai/character-packs) and fetches the @5dive/openagent package from the npm registry.
[REMOTE_CODE_EXECUTION]: The skill uses npx to execute the @5dive/openagent CLI tool and specifically mentions running code directly from the main branch of a GitHub repository via npx github:5dive-ai/openagent.
[PROMPT_INJECTION]: The skill processes external persona files that contain behavior and voice instructions. If sourced from untrusted third parties, these could influence agent behavior.
Ingestion points: The npx and 5dive agent import commands read data from <id>.persona.yaml files.
Boundary markers: Schema validation is performed, but no explicit instruction-isolation delimiters are mentioned for the persona content.
Capability inventory: The skill has access to CLI execution, media rendering, and system-level agent provisioning.
Sanitization: Input is validated against the OpenAgent v0.2 identity specification schema.

openagent