A penetration testing MCP server providing reconnaissance tools for security assessments including port scanning, subdomain enumeration, DNS enumeration, site mapping, vulnerability checking, and web technology detection.
Server has 18 tools with significant quality gaps. Most tool descriptions are present but lack LLM-optimization guidance (many are 50-150 chars, acceptable range but generic). Critical issue: 9 tools (find_suspicious_forms, check_csrf_tokens, check_cookies, site_mapper, check_vulnerabilities, fingerprint_technologies, ingest2DB, chunk_and_ingest, queryDB) accept Selenium WebDriver or complex object parameters with NO visible type definitions or schema constraints in the source. Input schemas are either missing or inferred. No tool has documented output schemas. Error handling is absent, no recovery guidance, no categorization. Naming is inconsistent: 'portScanner4LLM', 'subDomainMapper4LLM' use camelCase with '4LLM' suffix (non-standard); others use snake_case. No parameter descriptions for critical inputs like 'ports' range format or 'headers_to_check' array structure. Security concern: store_recon_results and ingest_results_to_db accept arbitrary JSON without validation. No idempotency hints, no permission gates, no audit trail declarations.
Check cookie attributes (Secure, HttpOnly) using Selenium WebDriver.
Check for CSRF tokens in forms on a webpage using Selenium WebDriver.
Check for verbose error messages by triggering potential errors on the target URL.
Check HTTP headers for presence of specified security headers.
Check for server information headers including Server and X-Powered-By.
Check TLS configuration for cryptographic failures including weak protocols and ciphers.
Check for vulnerabilities based on VulnerabilityList.json patterns in URL parameters, HTML content, headers, cookies, and JavaScript.
9 tools accept Selenium WebDriver or complex object parameters with NO visible type definitions or JSON Schema constraints. find_suspicious_forms, check_csrf_tokens, check_cookies, site_mapper, ingest2DB, chunk_and_ingest lack input schemas entirely.
No tool has a documented output schema. LLMs cannot plan downstream calls or extract required fields. store_recon_results, ingest_results_to_db, and queryDB return unspecified structures.
Tool naming is inconsistent and non-standard. portScanner4LLM, subDomainMapper4LLM, dnsEnumerator4LLM use camelCase with '4LLM' suffix instead of verb_noun convention (e.g., scan_ports, enumerate_subdomains). LLMs struggle to parse intent from non-standard names.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | <=2025-11-25 | v2 |
Chunk large reconnaissance data and ingest into the database.
Performs DNS enumeration including A, AAAA, MX, NS, TXT, SOA, and SRV record lookups for a target domain.
Find forms with input names that could indicate vulnerabilities using Selenium WebDriver.
Detect technologies using Wappalyzer patterns on rendered content including URL, HTML, headers, cookies, and JavaScript.
Ingest reconnaissance data into the database.
Ingest structured reconnaissance results to the database with source metadata.
Performs comprehensive port scanning on a target host including service detection, SSL/TLS analysis, OS fingerprinting, CDN detection, and web technology identification.
Query the reconnaissance database for stored results.
Recursively crawl a website and map its pages using Selenium for rendering and vulnerability checking.
Stores structured reconnaissance results (JSON) in the 'Reconnaissance' DB collection with source metadata.
Performs comprehensive subdomain enumeration using multiple techniques including brute-force, DNS queries, web search, and certificate transparency (crt.sh) lookups.
No error handling or recovery guidance. Tools like check_tls_configuration, check_error_disclosure, and queryDB provide no actionable error messages, categorization (retryable vs fatal), or next-step hints for LLM recovery.
store_recon_results and ingest_results_to_db accept arbitrary JSON objects without validation, type constraints, or schema documentation. No permission gates, audit trail declarations, or secret injection patterns. Security risk for unvalidated data ingestion.