AI session report
Generate a senior-engineer write-up of the session in seconds — root cause, evidence, fix recommendation. Optional, runs through your own API key.
A diagnostic console for browser-based softphones and UCaaS platforms. Drop a HAR, get a senior-engineer-grade verdict on calls, WebRTC, SIP and queue state — in seconds.
Telecom·HAR groups requests into sessions, reconstructs SIP dialogs, decodes SSE event streams and pulls every signal that explains a failed call.
Generate a senior-engineer write-up of the session in seconds — root cause, evidence, fix recommendation. Optional, runs through your own API key.
Reconstructs every dialog from dialog-info events. Detects stuck presence, terminated-without-identity bugs, zombie sessions.
ICE candidates, TURN allocation, DTLS, MOS estimate.
Agent state, wrap-up timing, missed-call attribution.
HID claim conflicts, mute mismatches, call-control desync.
Drop two HARs to diff timings, errors and dialog outcomes.
Token lifecycle, refresh windows, expiry collisions with calls.
CSP, HSTS, CORP, COEP, COOP — graded per-domain with fixes.
Delivery receipts, MMS attachments, conversation rebuild.
In the user's browser, open Network → enable "Preserve log" → reproduce the issue → right-click → Save all as HAR with content.
Drag the .har onto the analyser. The file never leaves the device. Even very large captures (500MB+) load with the optional body-stripping mode.
Sidebar badges flag where the issues are. Open the AI report for the executive summary, or dig into per-dialog SIP and WebRTC traces.
Built for support engineers handling real customer captures. Zero servers, zero accounts, zero audit risk.
HAR files are parsed entirely in your browser. Zero network requests leave this page during analysis.
No analytics, no third-party scripts. Settings persist in localStorage on this device only.
Save the .html file once, run it on an air-gapped machine forever. No external dependencies at runtime.
One static HTML file, no build step. Read every line of source before you trust it with customer data.
The fastest path from "the call dropped again" to "here's the line where it failed."