PcapAnalyzer is a free, Apache-2.0-licensed, self-hosted tool for diagnosing network problems from PCAP or PCAPNG captures, including gzip files. Its 27 detectors examine TCP behavior, routing and packet sizing, DNS and DHCP, application responses and TLS, tunnels, and security risks. They report a severity and suggested next step for issues such as retransmissions, DNS failures, PMTU black holes, slow HTTP responses, tunnel handshakes that fail, port scans, and insecure protocols. It also checks whether capture limitations - such as one-way visibility, truncated packets, or traffic mirrored more than once - could undermine the findings.
The interface offers Wireshark-style packet filtering and coloring, CSV export, protocol tables, response-time analysis, conversation flow diagrams, and PDF reports. Optional AI investigation can answer follow-up questions using packet-analysis tools and cite supporting packets; it requires the user’s Anthropic API key, though a no-cost demo mode is available. The project says captures are analyzed locally, with no telemetry or accounts in the self-hosted version. For AI sessions, capture data is encrypted and deleted after 60 minutes of inactivity or on request; the AI receives tool results rather than the PCAP itself, unless packet content is enabled for that session. Setup uses Docker Compose or a manual Python, Node, and TShark installation. The software is in beta, tested with synthetic and public captures, so its findings should be evaluated and feedback is encouraged.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.