The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches documentation from the vendor's domain using curl https://docs.reticle.sh/llms.txt. This resource is used to provide the agent with additional tool specifications and troubleshooting information.
[COMMAND_EXECUTION]: The skill provides instructions to run shell commands to initialize the vendor's environment, such as npx @reticlehq/server@latest init. It also suggests starting project-specific dev scripts from package.json to ensure the application is running during verification.
[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it ingests and processes data from a running web application which could contain untrusted content.
Ingestion points: The agent reads DOM snapshots, network request/response data, and console logs from the target web application using tools like reticle_snapshot and reticle_act_and_wait.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential commands embedded within the application's UI or logs.
Capability inventory: The skill environment allows for shell command execution (via npx and dev scripts) and network operations.
Sanitization: There is no explicit sanitization or filtering logic described for the data retrieved from the running application before it is processed by the agent.

verify-ui-change