The Agent Skills Directory

[COMMAND_EXECUTION]: The skill heavily utilizes shell commands via the Playdrop CLI (playdrop) to perform tasks such as authentication, project initialization, asset searching, and project publishing.
Evidence includes commands like playdrop project init ., playdrop auth login, and playdrop project publish . spread across all reference files and SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the official Playdrop CLI tool from the NPM registry.
Evidence: npm install -g @playdrop/playdrop-cli in SKILL.md.
Note: This is a vendor-owned resource consistent with the skill's primary purpose.
[DATA_EXFILTRATION]: The skill includes functionality to capture gameplay proofs and publish local project directories to the Playdrop platform.
Evidence: playdrop project capture and playdrop project publish . in references/publish-and-listing.md.
Note: This behavior is documented as the core feature of the platform integration.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Playdrop catalog and published demos which could theoretically contain malicious instructions targeting the AI agent.
Ingestion points: playdrop detail <ref> --json, playdrop browse, and playdrop search commands in references/discovery.md and references/assets-and-generation.md.
Boundary markers: No specific boundary markers or 'ignore' instructions are defined for processing the JSON output of these commands.
Capability inventory: The agent has the capability to execute shell commands and write to the local file system (via playdrop project create).
Sanitization: No explicit sanitization or validation of the remote catalog data is mentioned before use in project creation or remixing.

playdrop