How to search for exploits¶
Exploits are ordinary Vulners documents in the exploit bulletin family, so you search for
them with search.query and a family filter.
Find exploits for a CVE¶
Every exploit is an ExploitBulletin (or a per-collection subclass). On top of the common
bulletin fields, individual exploit collections add their own — e.g. exploitdb exposes
exploit_type and verified, seebug exposes has_poc — and any field the SDK has not modelled
stays reachable via extra="allow".
Find exploits for a product¶
Iterate every match lazily¶
iter_query (async: aiter_query) yields rows one at a time, fetching pages of
page_size behind the scenes and stopping at the
10,000-document window:
Expose exploit search to an AI agent (MCP)¶
The SDK ships a Model Context Protocol server so an LLM agent can search exploits (and CVEs, advisories, audits) with live data. Install the extra and run it:
The server exposes tools including search_exploits, search_bulletins, get_bulletin,
cve_lookup, audit_software, audit_linux and smart_audit, each returning compact JSON.
See AGENTS.md for the full
agent-facing guide.
Hosted vs. bundled
vulners-mcp here is a minimal, self-hosted server (the core tools above). For a
fully managed, always-on endpoint, use the official hosted server at
mcp.vulners.com — no install required.