The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it retrieves and analyzes content from external websites via the browser/markdown tool.
Ingestion points: The skill ingests untrusted markdown data from third-party URLs when diagnosing visibility gaps in AI answers as described in SKILL.md and references/geo-methodology.md.
Boundary markers: The instructions do not specify any boundary markers or delimiters to protect against instructions embedded within the fetched external content.
Capability inventory: The skill utilizes the unifapi MCP server for data retrieval. It does not possess capabilities for persistent storage, local file modification, or shell command execution.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from external sources before it is analyzed by the agent.
[EXTERNAL_DOWNLOADS]: The skill provides installation instructions for the unifapi-agent/agents plugin and the unifapi MCP server. These are official vendor resources and represent standard installation procedures.

ai-visibility-audit