agent-data-build
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external, untrusted websites which could contain malicious instructions.
- Ingestion points: Untrusted website content is fetched via the
agent-data parsecommand using theSOURCE_URLvariable inSKILL.md. - Boundary markers: There are no instructions to use delimiters or ignore instructions that may be found within the retrieved web content.
- Capability inventory: The skill executes shell commands (
agent-dataCLI), writes to local files (api-task.md), and makes network requests. - Sanitization: The instructions do not specify any validation or sanitization steps for the data retrieved from the
SOURCE_URLbefore the agent acts on it. - [COMMAND_EXECUTION]: The skill heavily utilizes shell commands to interact with the
agent-dataCLI tool. - Evidence: Commands like
agent-data parse "$SOURCE_URL" --task "$(cat api-task.md)"andagent-data parse eventsare used throughout the skill instructions.
Audit Metadata