architect-explore-agent
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
curlcommands to interact with the ElevenLabs API athttps://api.elevenlabs.io. These operations are used to fetch agent configuration, tools, procedures, and conversation metadata using an API key provided by the user. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and summarize external data, including agent system prompts and conversation transcripts, which represents a potential surface for indirect prompt injection if the data contains malicious instructions.
- Ingestion points: The skill fetches data from
/v1/convai/agentsand/v1/convai/conversations(SKILL.md). - Boundary markers: None specified; the skill does not explicitly require the use of delimiters when processing the fetched content.
- Capability inventory: The skill is limited to network reads via
curlagainst the vendor's official API. It does not have capabilities for writing to the file system or executing arbitrary shell commands beyond the defined API calls. - Sanitization: The skill focuses on summarization and does not describe explicit sanitization or filtering of the fetched content.
Audit Metadata