hunt-nextjs
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill performs active network reconnaissance and vulnerability testing by executing
curlcommands against a user-specified target variable ($TARGET). These commands are used to fingerprint versions, check for Server Action IDs, and test for SSRF or authentication bypasses. - [INDIRECT_PROMPT_INJECTION]: The skill fetches HTML and JSON content from remote servers and processes it using local Python scripts. This creates a potential attack surface where a malicious target could serve specially crafted content to influence the parsing logic.
- Ingestion points: Remote content is fetched from
$TARGETviacurlin multiple phases (e.g., Phase 5 and Phase 8). - Boundary markers: There are no explicit boundary markers or "ignore instructions" warnings used when the agent processes the output of these network requests.
- Capability inventory: The skill possesses network communication capabilities (
curl) and data processing capabilities (python3,grep). - Sanitization: The skill uses standard Python libraries (
json,re) to parse the ingested data but does not perform separate validation or sanitization of the remote content before parsing.
Audit Metadata