http-load-profiler

Pass

Audited by Gen Agent Trust Hub on Nov 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/http_benchmark.py executes external benchmarking tools (wrk and ab) using subprocess.run. The implementation safely passes arguments as a list and performs validation on the input URL to ensure it uses the http or https scheme and contains a valid hostname. This prevents shell command injection through maliciously crafted URLs.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from external URLs by performing load tests and reporting the results back to the agent. This creates a surface where an attacker-controlled endpoint could potentially influence the agent's behavior via the benchmarking output.
  • Ingestion points: Target url parameter in scripts/http_benchmark.py which is passed to external binaries for scanning.
  • Boundary markers: No explicit delimiters or 'ignore instructions' warnings are wrapped around the tool output before it is returned to the agent context.
  • Capability inventory: The script has the capability to execute system binaries via subprocess.run and parse their output.
  • Sanitization: Input URLs are validated for scheme and hostname before execution. The benchmarking tool output is filtered through specific regular expressions to extract only numeric performance metrics (RPS, latency percentiles, error rates), which significantly limits the potential for instruction injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 22, 2026, 01:11 AM