aegis-web-client
Fail
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: HIGHINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to fetch and process untrusted data from external web targets, creating a vulnerability surface for indirect prompt injection.
- Ingestion points: Multiple shell commands in the 'Exploitation & Reproduction' section of SKILL.md ingest data from app.example.com using curl.
- Boundary markers: The instructions employ grep to isolate specific markers within the response, but lack formal security boundaries or 'ignore' instructions for the rest of the external content.
- Capability inventory: The skill relies on shell command execution, network access via curl, and string processing via grep.
- Sanitization: There are no instructions for the agent to sanitize or validate the content retrieved from external sources before it is analyzed or reported.
- [COMMAND_EXECUTION]: The skill contains templates for shell commands that the agent is expected to execute to perform security testing. While these are documented using the placeholder domain app.example.com, they involve network operations and piping of remote data into local utilities (grep). These patterns were flagged by automated tools due to their structure, although their immediate purpose is instructional verification of web vulnerabilities.
Recommendations
- HIGH: Downloads and executes remote code from: https://app.example.com/login, https://app.example.com/profile, https://app.example.com/profile?name= - DO NOT USE without thorough review
Audit Metadata