Labs · things I run in public

Agents you can use today, with the bill attached

Each one is built on public data and self-hosted. Every query is priced in the open, because the point is to show what the public-data share of a six-figure subscription costs when an agent does the work. Decision support, not a replacement for your own programme.

Live

Vulnerability intelligence agent

Type a CVE, a product or a nickname. It pulls NVD, CISA KEV and FIRST EPSS, computes a P1–P4 priority, and writes the remediation guidance. Deep sweep adds live web search for public exploits and campaigns. Exports to Excel.

29lookups served

$0.004a fast lookup, about 11 s

$0.27a deep sweep, about 2 min

The ledger on the agent page lists every query's cost and turnaround. For contrast, mid-market threat-intel platforms run USD 100–250k a year. Final prioritisation still needs asset criticality from your CMDB.

Open the agent Read the ledger

Live

MCP server

The same agent as four tools any MCP client can call: Claude Code, Cursor, your own agent. Hosted here at https://clawofrohan.com/mcp, or as a token-free package you run yourself that never touches my keys.

claude mcp add vuln-intel -- uvx vuln-intel-mcp

vuln_lookupNVD + KEV + EPSS for one CVE, with priority

vuln_deep_sweepPublic exploit and campaign intel via web search

vuln_resolveProduct or nickname to candidate CVEs

vuln_ledgerWhat the last n calls cost and took

The package on PyPI Source on GitHub

Live · free

Weekly fitment

Five real requirements a week. Call each one OOB, config, custom or another tool, then read the model's call and reasoning. A share grid and the archive of past sets.

This week's set

Winding down

The Park, the RegTech console, the resume and profile builders

Experiments from earlier in 2026. They are being retired through September and October so the things that stay get the attention. Source is archived; if one of them mattered to you, tell me.