subdomain-takeover
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides the agent with instructions and specific shell commands for performing DNS lookups (e.g.,
dig) and active resource exploitation on cloud platforms (e.g.,aws s3 mb,heroku create). These commands are relevant to the skill's stated purpose of subdomain takeover detection and management. - [INDIRECT_PROMPT_INJECTION]: The playbook methodology requires the agent to ingest and analyze external data, such as DNS records and HTTP response bodies, for specific fingerprints. This creates a potential surface for indirect prompt injection if an attacker-controlled resource returns malicious instructions, although the risk is mitigated by the use of defined string-matching patterns.
- Ingestion points: DNS lookup results and HTTP response content from
curloperations. - Boundary markers: The instructions do not define specific delimiters to isolate external content from the agent's command logic.
- Capability inventory: The skill utilizes network scanning tools (
curl,dig) and administrative CLI tools for various cloud providers (aws,heroku,git). - Sanitization: The methodology relies on identifying specific service fingerprints rather than active input sanitization.
Audit Metadata