Five tools with basic descriptions and minimal parameter documentation. All tools are READ_ONLY reconnaissance functions with simple string/list inputs. Descriptions are present but generic (10-50 chars), lacking context on WHEN to use each tool or HOW results chain together. No output schemas documented. Parameters lack type constraints (enums, ranges, formats). No error handling guidance. Tool names follow verb_noun pattern (dns_enumerate, whois_lookup, crt_sh_certificates, wayback_snapshots, github_leaked_secrets) but descriptions do not explain dependencies or result structure. No pagination, no result limits stated. Security implications of leaked-secrets tool not addressed.

• Add output schema documentation for each tool. Example: dns_enumerate_tool returns {records: [{type: string, name: string, value: string, ttl: int}], total_count: int}. This enables LLMs to extract and chain results.
• Expand tool descriptions to 100-150 characters explaining WHEN to use each tool. Example: 'Enumerate DNS records (A, MX, NS, TXT) for a domain. Use this first to discover subdomains and mail servers. Returns structured records suitable for further investigation via whois_lookup or crt_sh_certificates.'
• Add parameter constraints: record_types should be an enum (A, AAAA, MX, NS, TXT, SOA, CNAME); from_date and to_date should specify ISO 8601 format; query should have min/max length (e.g. 3-100 chars).
• Document result limits and pagination. Example: 'Returns up to 50 most recent snapshots. Use from_date and to_date to narrow results. If more than 50 exist, call again with adjusted date range.'
• Add error handling guidance. Example: 'If domain not found, try a parent domain (e.g. example.com instead of sub.example.com). If rate limited, wait 60 seconds before retrying.'