dsh-log-analyzer

Log files routinely stretch to tens of thousands of lines, and manually scanning for every ERROR entry is both slow and prone to misses. Worse, the same underlying exception can masquerade as dozens of distinct failures simply because timestamps, request IDs, or file paths differ on each occurrence. dsh-log-analyzer solves exactly this problem: it ingests raw log text, recognises common error keywords such as ERROR, FATAL, Exception, Traceback, panic, and failed, extracts each error line with N lines of surrounding context, then groups them by a normalised signature—automatically stripping noise like timestamps, random numbers, and path fragments—to produce a concise, structured distribution of error types.

The plugin exposes two tools. analyze_log provides a bird's-eye view: it tallies the frequency of each error class, lists representative lines, and presents an overall distribution. extract_errors is more surgical, letting you supply a custom regex to target a specific pattern and retrieve every match alongside its context. For very large files, only the tail (default 50,000 lines) is read to keep things fast, and the window can be widened or narrowed via the maxLines option. Everything runs locally in pure Node—no network calls, no external services, and your log data never leaves the machine.

If you routinely debug production incidents, post-mortem crash dumps, or want your AI assistant to grasp a log before diagnosing it, this plugin compresses a wall of raw text into a clear error portrait, saving a great deal of manual filtering time.