agentaddress
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external sources (webhooks and emails) through the AgentAddress service. This untrusted content is then processed by the agent, creating a risk that malicious instructions could be embedded in the incoming data.
- Ingestion points: The agent fetches external data from the events endpoint described in SKILL.md.
- Boundary markers: The instructions lack specific guidance on using delimiters or ignore-instructions warnings to isolate untrusted external content.
- Capability inventory: The agent is instructed to use HTTP GET and POST methods to interact with the external API.
- Sanitization: There are no specified sanitization or validation steps for the content received from the inbox before it is introduced into the agent's context.
- [DATA_EXFILTRATION]: The skill facilitates network operations to the domain agentaddress.dev. While integral to the skill's purpose, communication with non-whitelisted external domains represents a risk surface for potential data exfiltration.
Audit Metadata