Kloudle

Teams that let AI agents build, test, and ship code can generate more commits in a night than a security team could review in a quarter. Traditional security review depends on a human reading a dashboard, which is already too slow when agents open pull requests while the team sleeps. Existing cloud security posture tools (CSPMs) also require centralizing security metadata, like asset inventories and privilege models, in the vendor's own cloud, which the site argues is a structural mismatch for teams that want to own their data. Meanwhile new attack surface is appearing in places point tools don't cover yet, like MCP servers and neocloud deployment targets such as Vercel, Railway, and Fly.

Kloudle runs 1,890 SQL-based security checks today across five providers (AWS, GCP, Azure, DigitalOcean, Kubernetes), with the checks readable rather than black-box scoring. It writes every issue to a PostgreSQL database the customer owns and self-hosts, so audit evidence for SOC 2 or PCI-DSS is queryable directly rather than held by a vendor. Any MCP-compatible agent can call search() and get() to pick up issues with remediation steps and work the same list as the human team. Scans run from a dashboard, CI, or cron, with exit codes (0 pass, 1 issues found, 2 errors) used to gate what ships. The company describes a broader roadmap of a 'posture layer' spanning cloud, MCP servers, apps and APIs, and agent-vs-human activity tracking, framed around the idea that humans set the gates once and agents work the list at volume.

The pitch is that agent-driven 'software factories' now ship code and open pull requests at machine speed, so security controls need to run inside that loop rather than depend on periodic human dashboard review.