The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from the application being debugged, which could potentially contain malicious instructions.
Ingestion points: Untrusted data enters the agent context through tools like reticle_console, reticle_network, reticle_state, and reticle_storage as described in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters mentioned to prevent the agent from potentially obeying instructions embedded within the application's logs or state.
Capability inventory: The skill possesses extensive inspection capabilities (element properties, network traffic, storage, and application state) but lacks arbitrary write access or non-vendor exfiltration tools.
Sanitization: The skill documentation does not mention specific sanitization or filtering of the ingested runtime data.
[EXTERNAL_DOWNLOADS]: The documentation provides instructions to install the Reticle server using npx @reticlehq/server@latest init. The package originates from the skill's official vendor.
[COMMAND_EXECUTION]: The skill instructs the user or agent to execute shell commands for setup and troubleshooting, including npx for the server installation and curl to fetch vendor-hosted documentation from docs.reticle.sh.

debug-broken-ui