Online HAR viewer: open .har files exported from browser DevTools — requests grouped by domain, seven-segment timing waterfalls, status distribution, slowest/largest request spotting. Fully local analysis, captures never leave your browser.
TL;DR: Online HAR viewer: domain grouping, seven-segment timing waterfalls, status distribution, local-only analysis.
For local tools, open DevTools → Network and search your input — it should not appear in requests.
Usage Guide
Inspect HAR locally for status, timing, and size — keep secrets out of uploads.
1. Use
Open a HAR, filter failures/slow calls, inspect headers and timing.
2. What to hunt
4xx/5xx, dupes, heavy payloads, third-party blocking, bad timing segments.
3. Safety
HAR often contains cookies/tokens — treat it as sensitive.
4. Perf reading
Separate server wait from download time; watch cache/dupes.
5. Related
Copy one call as curl → codegen/API tester for reproduction.
6. Accept
Ensure the recording captured the interaction you care about.
FAQ
Chrome/Edge: DevTools → Network → reproduce the flow → right-click the request list → "Save all as HAR with content". Firefox: Network panel gear menu → "Save All As HAR". Then open it here.
Yes — cookies, Authorization headers and request bodies are all in there. This tool parses locally with zero uploads, but sanitize a HAR before sharing it with others.
blocked (queueing) → dns → connect (TCP) → ssl (TLS) → send → wait (TTFB) → receive. Long gray/purple means slow connection setup; long blue wait means server-side latency; long teal receive means big payloads or low bandwidth.
In the HAR spec dns/connect/ssl may overlap (ssl happens inside connect), and cached requests report -1 for phases that never happened. The tool shows raw segment values — for bottlenecks, look at the longest segment.
Filter for failures (4xx/5xx) and slow requests first, then compare the waterfalls of the slow ones: a long wait segment means server-side latency, a long receive segment means a big payload or thin bandwidth. The list view also exposes duplicate requests and cache hits; when third-party domains dominate, treat blocking scripts as prime suspects.
Confirm the interaction you care about actually made it into the capture — incognito windows and browser extensions can change the request set, and analyzing the wrong scope is wasted effort no matter how thorough. To reproduce a single suspect request, copy it as cURL from DevTools: the HAR gives you the global view, per-request tools handle reproduction.
Related Tools
Related Articles
Harvest Now, Decrypt Later: Whose Data Needs Post-Quantum Crypto First
The HNDL attack model, taken apart: why Shor's algorithm kills RSA/ECC while AES-256 stays standing, how data lifetime sets migration priority, which systems can relax, and how Chinese cryptographic compliance (SM/TLCP) and PQC stack rather than compete.
HKDF: How TLS 1.3 Grows One Shared Secret Into a Whole Key Tree
A TLS handshake yields one shared secret (or two concatenated in a hybrid group), but a session needs dozens of keys: bidirectional write keys, IVs, per-stage Finished keys, exporters, resumption PSKs. HKDF (RFC 5869) does that job in Extract + Expand. This post covers structure, salt/info domain separation, boundaries with HMAC and PBKDF2, and common misuses.
ASCII vs Unicode vs UTF-8: Character Encoding Finally Explained
ASCII has 128 characters, Unicode covers 140k+, UTF-8 is Unicode's variable-length encoding. Learn the difference, byte vs character length, and why most encoding bugs happen.