The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill recommends a one-line installation method curl -fsSL https://raw.githubusercontent.com/abhi1693/openclaw-mission-control/master/install.sh | bash in the 'Quick Start with Installer' section of SKILL.md. This pattern pipes content from a remote server directly into a shell interpreter, which is a high-risk execution pattern.
[EXTERNAL_DOWNLOADS]: The skill instructs users to download and execute code from a repository at github.com/abhi1693. This source is an unverified individual account and does not belong to a trusted organization or the established vendor infrastructure for this skill.
[COMMAND_EXECUTION]: The documentation provides numerous shell commands for system configuration, service management, and troubleshooting (e.g., docker compose, lsof, chmod, npm install) which, when combined with instructions to download external scripts, increases the risk of unauthorized system modification.
[INDIRECT_PROMPT_INJECTION]: The skill implements a dashboard that ingests untrusted data which is later processed and displayed to agents and users, creating a potential injection surface.
Ingestion points: API endpoints for creating tasks (createTask) and requesting approvals (requestApproval) in SKILL.md accept user-provided title, description, and reason strings.
Boundary markers: Absent. The documentation does not specify the use of delimiters or instructions to ignore embedded commands within these fields.
Capability inventory: The skill manages agent assignments, distributed task execution via gateways, and sensitive approval workflows.
Sanitization: No sanitization or validation logic is evident in the provided integration examples for the data stored in the hierarchical task structure.

openclaw-mission-control