The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the duvo CLI via Bash to execute commands like duvo runs get and duvo runs messages to retrieve metadata and transcripts in JSON format.
[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests untrusted data from run transcripts and Agent Orientation Procedures (AOPs) provided by the Duvo API or user input. (1) Ingestion points: messages retrieved via listRunMessages, listCaseRunRecentMessages, and revision data from getRevision. (2) Boundary markers: The skill includes instructions to "quote, do not paraphrase" and "not to invent transcript content." (3) Capability inventory: The skill can execute CLI commands via Bash and read run/revision data. (4) Sanitization: There is no explicit sanitization or filtering logic described for the external content.
[EXTERNAL_DOWNLOADS]: The skill references resources from the vendor's official domains (duvo.ai, docs.duvo.ai), their public GitHub repository (github.com/duvoai/skills), and the official NPM registry for the @duvoai/cli package.

run-debugger