Servus from Austria!

Im the creator of Hecate, a free and open-source vulnerability management platform you can host yourself. Ive posted here months ago about Hecate and thought after few new features and bug fixes to post again here. It all started as the prototype for my masters thesis and ive been running it at work for several months now. The masters thesis was actually about using AI with CVE and if or how the analysis/work improves/worsens, will be faster/slower, etc…

The main reasons actually why ive built Hecate is because i wanted a vulnerability database that takes work off my shoulders instead of being one more feed to check. AI is a nice to have but the other core features around Hecate are way more useful. It should tell me when a new vuln hits something we actually run and warn me when a package in one of our repos turns out to be malware… centralized, no rate limits, no additional costs… i didnt want to piece that together from five different sites every morning.

Hecate pulls NVD, EUVD, CISA KEV, GHSA, OSV and CIRCL (with EPSS) into one database. You can search it by keyword, DQL or regex or just ask whether a specific version is affected. It also follows CERT, vendor and community advisory feeds (CERT-FR, CERT-EU, NCSC-NL, JPCERT/CC, Mastodon user RSS and so on) and groups advisories about the same CVEs. That way you notice quickly when several sources are talking about the same thing.

There are two ways to tell it what you run. The inventory is a list of products and versions. Hecate matches it against the advisories using version ranges and shows the support/EOL status from endoflife.date next to each entry. For your own or third party code and images theres SCA: it scans container images and git repos automatically and rescans when the digest or commit changes. Under the hood it uses Trivy, Grype, Syft, OSV-Scanner, Semgrep, TruffleHog, govulncheck and its own analyzer for malware heuristics. SBOMs (CycloneDX, SPDX) and VEX (CycloneDX, OpenVEX, CSAF) can be imported and exported… and can be mapped automatically on scans for ISO:27001 reasons.

The malware part was important to me. Hecate keeps the OSV malicious package records (400k+ across npm, PyPI and others). When a package gets reported as malicious after your last scan, its flagged in that scan right away instead of waiting for the next one. When a rule matches, notifications go out through Apprise (Slack, Teams, Signal, mail,… etc).

AI is opt-in. Everything above works without an API key but better with the free ones of course. If you want triage text, you can connect OpenAI, Anthropic, Gemini or a local model like Ollama.

At my work its connected to our ticketing system through automations: new matches and findings end up as tickets. Our repos get scanned from the pipelines through the REST API. It runs with Docker Compose, and the images are on GHCR.

Im looking specifically for testers. Set it up, use it, break it and tell me whats missing or annoying. GitHub issues or comments here both work… im especially interested in matching quality (false positives and negatives) and in how painful the setup is. Hecate is now about 13 months old and i plan to continue improving it. I know theres still a lot to do.

Full disclosure: I build Hecate with AI assistance. The design decisions, the testing and running it in production are on me.