The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (HIGH): The skill utilizes the npx skills add command to download and execute external code. This is a direct vector for Remote Code Execution (RCE) as it allows the agent to pull and install software from remote repositories.
[COMMAND_EXECUTION] (HIGH): The skill constructs and executes shell commands (npx skills find [query]) using unvalidated user input. This creates a surface for command injection where a user or an attacker-controlled input could append malicious shell operators.
[EXTERNAL_DOWNLOADS] (HIGH): The skill encourages downloading packages from both trusted and untrusted sources. While vercel-labs/agent-skills is a trusted source, the skill specifically suggests other untrusted sources like ComposioHQ/awesome-claude-skills and 'other sources', which circumvents security best practices for dependency management.
[INDIRECT PROMPT INJECTION] (HIGH): The skill processes external content (search results from the npx skills find command) and uses that content to suggest or perform further high-privilege actions like installation.
Ingestion points: Search results from the Skills CLI and user-provided queries.
Boundary markers: None present. The skill directly interpolates external strings into the agent's workflow.
Capability inventory: The skill has the ability to execute shell commands and modify the system via package installation.
Sanitization: No sanitization or validation of the search results or package identifiers is performed before installation.

find-skills