visual-content

When the user wants to plan, create, or repurpose visual content (images, infographics, social post images) across channels. Also use when the user mentions...

This skill looks like a normal content/visual planning helper and has no installers or credential requests, but its instructions tell the agent to read local project-context files (.claude/project-context.md and .cursor/project-context.md) even though the skill metadata doesn't declare those as required config paths. That means the agent could access workspace files you might not expect it to. Before installing or enabling: (1) Inspect those project-context files (or remove/sanitize them) if they contain any sensitive info; (2) ask the skill author to explicitly declare file access in the metadata (requires.configPaths) or to make file-reading conditional on explicit user permission; (3) run the skill in a sandbox or only on projects where those files are safe to share; and (4) if you need clarification, request a version of the skill that prompts the user before reading any local files. These steps will reduce the risk that the agent unintentionally reads sensitive project data.

Type: OpenClaw Skill Name: visual-content Version: 1.0.1 The 'visual-content' skill bundle is a purely instructional set of guidelines for an AI agent to assist users with visual content planning and image specifications. It contains no executable code, and its instructions (SKILL.md) are strictly aligned with its stated purpose of managing social media and website imagery without any signs of prompt injection or data exfiltration.