Download the PHP package alfism1/filament-log-management without Composer

Filament Log Management

A Filament 3 log viewer that filters Laravel log entries by the process that produced them — cron, queue, mail, database, and whatever domain processes you configure — on top of the usual level / text / date filters.

• Reads every *.log in storage/logs, newest entry first
• Groups entries into configurable processes, with hit counts in the filter dropdowns
• Full-text search across message, context and stack trace
• Rows expand to show the full message, pretty-printed context and the stack trace
• Unwraps Laravel exception entries, whose trace is buried inside the JSON exception key
• Auto-refresh, download, empty and delete actions
• Never loads a whole file: it streams backwards from the end under a scan budget
• English and Indonesian translations included

Requirements

Installation

Register the plugin on the panel:

That's it — the page appears under a Developer navigation group at /admin/log-management.

Publish the config to change anything:

Views and translations are publishable too, with --tag=filament-log-management-views and --tag=filament-log-management-translations.

Configuring the plugin

Everything is optional; anything left unset falls back to the config file.

Processes — the point of the package

Every entry is classified into the first matching process, so config order matters: put specific processes above general ones. A process matches when either

• files — the log file name matches one of its globs, or
• patterns — one of its regexes matches "<message> <context>".

An explicit [tag] prefix on the message always wins over both:

The tag is matched against every process key and its aliases. A tag matching nothing configured still gets its own filter option, labelled from the tag — so [midtrans-webhook] shows up as "Midtrans Webhook" with no config change at all.

Adding your own

The shipped defaults cover framework concerns only (scheduler, queue, notification, mail, upload, auth, database, http). Add your domain processes above them in the published config:

Domain nouns want that pattern pair: /\bword/i catches prose, /[a-z]Word/ catches camelCase identifiers quoted in the message. The camelCase form is also what stops stripe from matching a trip process.

Order tips learned the hard way: put invoice and payment above booking (a message mentioning both is usually about the more specific one), and a dedicated-channel process like livechat above the generic one it would otherwise fall into.

Scan depth

Log files grow without bound, so the file is never fully loaded. It is read backwards in 256 KB chunks and parsed newest-first, stopping at the requested depth (default 2,000 entries), the max_bytes cap, or the start of the file.

Filters therefore apply to the last N entries, not to the whole file. When the scan stops early the page says so and points at the depth selector. Raising the depth costs a linear scan — roughly 900 entries across 3.8 MB parses in ~150 ms.

Authorization

By default any user who can reach the panel can open the page. To gate it, either set a Gate ability in the config:

or pass a closure to the plugin, which overrides the config:

With Filament Shield

Shield discovers plugin-registered pages, so:

then set 'permission' => 'page_LogViewer', or use ->authorize(fn () => auth()->user()->can('page_LogViewer')). super_admin passes either way through Shield's Gate::before.

Config reference

Set LOG_MANAGEMENT_ALLOW_DELETE=false in production if log files should only ever be rotated by the system.

Under the hood

Testing

License

MIT. See LICENSE.md.