cloud-infra-detector

Warn

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill defines a PostToolUse hook that executes a shell script using a relative path traversal: ../../../hooks/skills/post_output_validation_hook.sh. This command attempts to run a script located multiple levels above the skill's own directory, which could allow for the execution of unauthorized code outside of the skill's sandbox environment.
  • [PROMPT_INJECTION]: The skill is designed to ingest and process raw data from external sources, including DNS records, HTTP headers, and IP signals, which provides a surface for indirect prompt injection attacks.
  • Ingestion points: Untrusted data enters the skill via the ip_signals, dns_signals, http_signals, tls_signals, and repository_signals inputs described in SKILL.md.
  • Boundary markers: There are no boundary markers or instructions provided to the agent to ignore embedded commands within the ingested data signals.
  • Capability inventory: The skill possesses the capability to read files, perform pattern matching with Grep, and execute shell scripts via the command hook.
  • Sanitization: The provided detection logic shows no evidence of sanitizing or validating input signals before they are used to determine infrastructure attribution or included in the final output JSON.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 9, 2026, 11:36 PM