# What is sent

> What dlogify receives, what it redacts and what it does not do.

This page lists what leaves your application, what dlogify removes on arrival, and who else sees it.

## What clients send

The dlogify clients send only log records: the message, level and time, the attributes you add, the error's type, message and stack trace, and the service name, environment, release, host name and client version. They make no other network calls, open no ports and collect nothing else about the machine.

## Personal data

dlogify redacts the message, the stack trace and every attribute value when it processes a batch, before anything is written to its database or sent to a third party. Until then, incoming batches wait briefly, unredacted, in an internal queue; a batch that cannot be processed is kept there for at most 14 days for investigation. Redaction replaces:

| Found | Replaced with |
|---|---|
| JSON Web Tokens | `<jwt>` |
| `Bearer` tokens | `Bearer <token>` |
| Keys with well-known prefixes (`sk_live_`, `sk_test_`, `rk_live_`, `ghp_`, `gho_`, `ghs_`, `github_pat_`, `glpat-`, `xoxb-`, `xoxp-`, `AKIA…`, `AIza…`) | `<secret>` |
| Values after `password`, `passwd`, `secret`, `token`, `api_key`, `apikey` or `access_key` followed by `=` or `:` | `<secret>` (the key name is kept) |
| Email addresses | `<email>` |
| Card numbers (13–19 digits that pass the Luhn check) | `<card>` |
| IPv4 and IPv6 addresses | `<ip>` |

Redaction is a safety net, not a guarantee: avoid logging personal data you do not need. To keep data from leaving your process at all, redact it in the client: see [Redaction](https://docs.dlogify.com/send-logs/redaction.md).

## AI providers

To classify an error and write its ticket, dlogify sends AI providers the redacted information about the error group: the message, the stack trace, the surrounding log lines and counts. INFO summaries use counts and a few redacted examples. Providers are used with settings that exclude your data from training. The list of providers is in the [privacy policy](https://dlogify.com/privacy).

## Retention

Error samples, tickets and INFO summaries are deleted after your plan's retention period (see [Plans and usage](https://docs.dlogify.com/dashboard/plans-and-usage.md#retention)).

## Log history

dlogify stores the redacted lines that reach it, of every level, so you can [replay](https://docs.dlogify.com/dashboard/replay.md) what led to a ticket: time, level, service, environment, message and trace ID, with the message cut at 2 KiB. Stack traces and attributes are not part of it. Lines are kept for your plan's retention on Cloudflare R2 in the United States, and an admin can turn this off [per project](https://docs.dlogify.com/dashboard/projects-and-keys.md#log-history).

SDKs default to `minLevel: info`, so `trace` and `debug` lines are stored only if you send them. Set a lower `minLevel` in the SDK to have them in the replay; they count toward your plan's events.

## What dlogify does not do

- It does not group `INFO`, `trace` and `debug` events: `INFO` is counted and summarized, and `trace` and `debug` never become tickets (see [Level mapping](https://docs.dlogify.com/send-logs/limits.md#level-mapping)).
- It does not call your systems, except the webhooks you configure.

Source: https://docs.dlogify.com/send-logs/what-is-sent
