The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent or user to download and execute a shell script directly from the vendor's website.
Evidence: bash <(curl -fsSL https://payram.com/setup_payram.sh) --testnet in SKILL.md.
Context: This is a vendor-owned resource (payram.com) used for the primary purpose of the skill (installation).
[COMMAND_EXECUTION]: The skill requires the execution of several system-level commands to configure the host environment and verify deployment.
Evidence: Use of docker port to verify network bindings and health checks.
[PRIVILEGE_ESCALATION]: The skill commands require elevated permissions to modify the system's security posture and install software.
Evidence: Execution of sudo ufw to modify firewall rules and the requirement to SSH as a root or sudo user.
[INDIRECT_PROMPT_INJECTION]: The skill defines an interface for MCP tools that process user-supplied data which could be used to inject malicious instructions if not properly handled.
Ingestion points: Parameters in the payram_setup_plan tool call within SKILL.md (e.g., domain, integration type).
Boundary markers: No explicit delimiters or instructions to ignore embedded content are provided in the documentation.
Capability inventory: Access to system shell, network utilities, and sensitive file system paths (e.g., ~/.payraminfo/aes/).
Sanitization: The skill does not describe any sanitization or validation logic for the external input data it processes.

payram-setup