vibe-pentest-ai-security-testing

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly extracts session cookies, local/session storage and Authorization headers and shows using them in subprocess/CLI calls (e.g., --cookies "session=abc123") and HTTP requests, which requires the agent to handle and potentially emit secret values verbatim.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most links point to reputable/docs (bilibili, projectdiscovery/katana, example.com, ara.so), but the ok-helloworld/vibe-pentest GitHub and its Gitee mirror are personal/unvetted repositories that may bundle or point to third‑party binaries (including crawler tools) and thus represent a potentially risky download source.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). At runtime, the workflow crawls and reads outsider-authored web content from the target site via Katana (scripts/run_katana.py → workspace/crawl_summary.json / crawled_anonymous.jsonl), and then feeds the discovered endpoints/parameters into the multi-agent LLM context through scripts/prepare_agent_findings.py and subsequent evidence/report generation.