The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references fetching documentation and configuration from the vendor's official site at docs.reticle.sh. It also describes an installation procedure for the core component using the vendor's package registry.
[COMMAND_EXECUTION]: The documentation provides an initialization command using npx @reticlehq/server@latest init to set up the auditing environment.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests and processes data from live web applications which may contain untrusted content.
Ingestion points: Data retrieved via reticle_capabilities and reticle_run, including DOM element IDs, console logs, and network request/response data from the target application (SKILL.md).
Boundary markers: None explicitly defined in the skill instructions to separate application data from agent instructions.
Capability inventory: The agent uses reticle_run to execute diagnostic tools like reticle_verify and reticle_explore (SKILL.md).
Sanitization: There are no instructions for sanitizing or filtering the content harvested from the web application before processing.

audit-my-app