> For the complete documentation index, see [llms.txt](https://docs.panther.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.panther.com/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types.md).
# Custom Logs
Define, write, and manage custom schemas
## Overview
Panther allows you to define your own custom schemas. You can ingest custom logs into Panther via a [Data Transport](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports.md), and your custom schema will then normalize and classify the data.
This page explains how to define, write, and manage custom schemas, as well as how to upload schemas with [Panther Analysis Tool (PAT)](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/panther-developer-workflows/ci-cd/deployment-workflows/pat.md). For information on how to use `pantherlog` to work with custom schemas, please see [`pantherlog` CLI tool](#using-pantherlog-cli).
Custom schemas are identified by a `Custom.` prefix in their name and can be used wherever a natively supported log type is used:
* Log ingestion
* You can onboard custom logs through a [Data Transport](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports.md) (e.g., HTTP webhook, S3, SQS, Google Cloud Storage, Azure Blob Storage)
* Detections
* You can write [rules and scheduled rules](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/detections/rules.md) for custom schemas.
* Investigations
* You can query the data in [Search](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/search/search-tool.md) and in [Data Explorer](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/search/data-explorer.md). Panther will create a new table for the custom schema once you onboard a source that uses it.
## How to define a custom schema
{% hint style="info" %}
Panther supports JSON data formats and CSV with or without headers for custom log types. For *inferring* schemas, Panther does not support CSV without headers.
{% endhint %}
There are multiple ways to define a custom schema. You can:
* Infer one or more schemas from data:
* In the Panther Console:
* To infer a schema from sample data you've uploaded, see the [Inferring a custom schema from sample logs](#inferring-a-custom-schema-from-sample-logs) tab, below.
* To infer a schema from S3 data received in Panther, see the [Inferring a custom schema from S3 data received in Panther](#inferring-a-custom-schema-from-s3-data-received-in-panther) tab, below.
* To infer one or more schemas from historical S3 data, see the [Inferring custom schemas from historical S3 data](#inferring-custom-schemas-from-historical-s3-data) tab, below.
* To infer a schema from HTTP data received in Panther, see the [Inferring a custom schema from HTTP data received in Panther](#inferring-a-custom-schema-from-http-data-received-in-panther) tab, below.
* In the CLI workflow:
* Use the [`pantherlog infer`](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/panther-developer-workflows/pantherlog.md#infer-generate-a-schema-from-json-log-samples) command
* Create a schema manually:
* See the [Creating a custom schema manually](#creating-a-custom-schema-manually) tab, below.
{% tabs %}
{% tab title="Sample logs" %}
## Inferring a custom schema from sample logs
You can generate a schema by uploading sample logs into the Panther Console. If you'd like to use the command line instead, follow the [instructions on using the pantherlog CLI tool here](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/panther-developer-workflows/pantherlog.md#generating-a-schema-from-json-samples).
To get started, follow these steps:
1. Log in to your Panther Console.
2. On the left sidebar, navigate to **Configure > Schemas.**
3. At the top right of the page next to the search bar, click **Create New**.
4. Enter a **Schema ID**, **Description**, and **Reference URL**.
* The Description is meant for content about the table, while the Reference URL can be used to link to internal resources.
5. Optionally enable **Field Discovery** by clicking its toggle `ON`. Learn more in [Enabling field discovery](#enabling-field-discovery).
6. Scroll to the bottom of the page where you'll find the option to upload sample log files.
7. Upload a sample set of logs: Drag a file from your computer over the "Infer schema from sample logs" box or click **Select file** and choose the log file. Note that Panther does not support CSV without headers for inferring schemas.\
* After uploading a file, Panther will display the raw logs in the UI. You can expand the log lines to view the entire raw log. Note that if you add another sample set, it will override the previously-uploaded sample.
8. Select the appropriate **Stream Type** ([view examples for each type here](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/reference.md#stream-type)).
* **Lines:** Events are separated by a new line character.
* **JSON:** Events are in JSON format.
* **JSON Array:** Events are inside an array of JSON objects.
* **CloudWatch Logs:** Events came from CloudWatch Logs.
9. Click **Infer Schema**
* Panther will begin to infer a schema from the raw sample logs.
* Panther will attempt to infer multiple timestamp formats.
* Once the schema is generated, it will appear in the schema editor box above the raw logs.\
10. To ensure the schema works properly against the sample logs you uploaded and against any changes you make to the schema, click **Validate & Test Schema.**
* This test will validate that the syntax of your schema is correct and that the log samples you have uploaded into Panther are successfully matching against the schema. You should see the results appear below the schema editor box.
* All successfully matched logs will appear under **Matched;** each log will display the column, field, and JSON view.\
* All unsuccessfully matched logs will appear under **Unmatched;** each log will display the error message and the raw log.
11. Click **Save** to publish the schema.
{% hint style="info" %}
Panther will infer from all logs uploaded, but will only display up to 100 logs to ensure fast response time when generating a schema.
{% endhint %}
{% endtab %}
{% tab title="S3 data received in Panther" %}
## Inferring a custom schema from S3 data received in Panther
You can generate and publish a schema for a custom log source from live data streaming from an S3 bucket into Panther. You will first [view your S3 data](#view-raw-s3-data) in Panther, then [infer a schema](#infer-a-schema-from-raw-data), then [test the schema](#test-the-schema-with-raw-data).
### **View raw S3 data**
After onboarding your S3 bucket into Panther, you can view raw data coming into Panther and infer a schema from it:
1. Follow the instructions to [onboard an S3 bucket onto Panther](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/aws/s3.md) without having a schema in place.
2. While viewing your log source's **Overview** tab, scroll down to the **Attach a schema to start classifying data** section.\
3. Choose from the following options:
* **I want to add an existing schema:** Choose this option if you already created a schema and you know the S3 prefix you want Panther to read logs from. Click **Start** in the tile.
* You will see a **S3 Prefixes & Schemas** popup modal:\
* **I want to generate a schema from raw events:** Select this option to generate a schema from live data in this bucket and define which prefixes you want Panther to read logs from. Click **Start** in the tile.
* Note that you may need to wait up to 15 minutes for data to start streaming into Panther.
* On the page you are directed to, you can view the raw data Panther has received at the bottom of the screen:\
* This data is displayed from `data-archiver`, a Panther-managed S3 bucket that retains raw logs for up to 15 days for every S3 log source.
* Only raw log events that were placed in the S3 bucket *after* you configured the source in Panther will be visible, even if you've set the timespan to look further back.
* If your raw events are JSON-formatted, you can view them as JSON by clicking **View JSON** in the left-hand column.
### **Infer a schema from raw data**
If you chose to **I want to generate a schema from raw events** in the previous section, now you can infer a schema.
1. Once you see data populating in **Raw Events,** you can filter the events you'd like to infer a schema from by using the string Search, S3 Prefix, Excluded Prefix, and/or Time Period filters at the top of the **Raw Events** section.
2. Click **Infer Schema** to generate a schema.\
3. On the **Infer New Schema** modal that pops up, enter the following:
* **New Schema Name:** The name of the schema that will map to the table in the data lake once the schema is published.
* The name will always start with `Custom.` and must have a capital letter after.
* **S3 Prefix:** Use an existing prefix that was set up prior to inferring the schema or a new prefix.
* The prefix you choose will filter data from the corresponding prefix in the S3 bucket to the schema you've inferred.
* If you don't need to specify a specific prefix, you can leave this field empty to use the catch-all prefix that is called `*`.\
4. Click **Infer Schema**.
* At the top of the page, you will see **'\' was successfully inferred**.
* Click **Done**.\
* The schema will then be placed in a **Draft** mode until you're ready to publish to production after testing.
5. Review the schema and its fields by clicking its name.\
* Since the schema is in **Draft**, you can change, remove, or add fields as needed.\
### **Test the schema with raw data**
Once your schemas and prefixes are defined, you can proceed to testing the schema configuration against raw data.
1. In the **Test Schemas** section at the top of the screen, click **Run Test**. \
2. On the **Test Schemas** modal that pops up, select the **Time Period** you would like to test your schema against, then click **Start Test**.\
* Depending on the time range and amount of data, the test may take a few minutes to complete.\
* Once the test is started, the results appear with the amount of matched and unmatched events.
* **Matched Events** represent the number of events that would successfully classify against the schema configuration.
* **Unmatched Events** represent the number of events that would not classify against the schema.
3. If there are **Unmatched Events**, inspect the errors and the JSON to decipher what caused the failures.\
* Click **Back to Schemas**, make changes as needed, and test the schema again.
4. Click **Back to Schemas**.
5. In the upper right corner, click **Save**.\
* The inferred schema is now attached to your log source.
{% endtab %}
{% tab title="Historical S3 data" %}
## Inferring custom schemas from historical S3 data
You can infer and save one or multiple schemas for a custom S3 log source from historical data in your S3 bucket (i.e., data that was added to the bucket *before* it was onboarded as a log source in Panther).
### Prerequisite: Onboard your S3 bucket to Panther
* Follow the instructions to [onboard an S3 bucket onto Panther](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/aws/s3.md) without having a schema in place.
* If you have onboarded the S3 source [with a custom IAM role](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/aws/s3.md#i-want-to-set-everything-up-on-my-own), that role must have the `ListBucket` permission.
### **Step 1: View the S3 bucket structure in Panther**
After creating your S3 bucket source in Panther, you can view your S3 bucket's structure and data in the Panther Console:
1. In the Panther Console, navigate to **Configure > Log Sources**. Click into your S3 log source.
2. In the log source's **Overview** tab, scroll down to the **Attach a Schema to start classifying the data** section.
3. On the right side of the **I want to generate a schema from bucket data** tile, click **Start**.
* You will be redirected to a folder inspection of your S3 bucket. Here, you can view and navigate through all folders and objects in the S3 bucke.
* Alternatively, you can access the folder inspection of your S3 bucket via the success page after [onboarding your S3 source](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/aws/s3.md) in Panther. From that page, click **Attach or Infer Schemas**.\
### **Step 2: Navigate through your data**
* While viewing the folder inspection, click an object.
* A preview window will appear, displaying a preview of its events:
If the events fail to render correctly (either generating an error or displaying events improperly), it's possible the wrong stream type has been chosen for the S3 bucket source. If this is the case, click **Selected Logs Format is n**:
### **Step 3: Indicate if each folder has existing schema or a new one should be inferred**
After reviewing what's included in your bucket, you can determine if one or multiple schemas is necessary to represent all of the bucket's data. Next, you can select folders that include data with distinct structures and either infer a new schema, or assign an existing one.
1. Determine whether one or more schemas will need to be inferred from the data in your S3 bucket.
* If all data in the S3 bucket is of the same structure (and therefore can be represented by one schema), you can leave the default **Infer New Schema** option selected on the bucket level. This generates a single schema for all data in the bucket.
* If the S3 bucket includes data that need to be classified in multiple schemas, follow the steps below for each folder in the bucket:
1. Select a folder and click **Include**.
* Alternatively, if there is a folder or subfolder that you do *not* want Panther to process, select it and click **Exclude**.\
2. If you have an existing schema that matches the data, click the **Schema** dropdown on the right side of the row, then select the schema:\
* By default, each newly included folder has the **Infer New Schema** option selected.
2. Click **Infer `n` Schemas**.\
### Step 4: Wait for schemas to be inferred
The schema inference process may take up to 15 minutes. You can leave this page while the process completes. You can also stop this process early, and keep the schema(s) inferred during the time that the process ran.
### Step 5: Review the results
After the inference process is complete, you can view the resulting schemas and the number of events that were used during each schema's inference. You can also validate how each schema parses raw events.
1. Click the play icon on the right side of each row.
2. Click the **Events** tab to see the raw and normalized events.
3. Click the **Schema** tab to see the generated schema.
### Step 6: Name the schema(s) and save source
Before saving the source, name each of the newly inferred schemas with a unique name by clicking **Add name**.
After all new schemas have been named, you will be able to click **Save Source** in the upper right corner.
{% endtab %}
{% tab title="HTTP data received in Panther" %}
## Inferring a custom schema from HTTP data received in Panther
You can generate and publish a schema for a custom log source from live data streaming from an HTTP (webhook) source into Panther. You will first [view your HTTP data](#view-raw-http-data) in Panther, then [infer a schema](#infer-a-schema-from-raw-data-1), then [test the schema](#test-the-schema-with-raw-data-1).
### **View raw HTTP data**
After creating your [HTTP source](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/http.md) in Panther, you can view raw data coming into Panther and infer a schema from it:
1. Follow the [instructions to set up an HTTP log source](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/http.md#how-to-set-up-an-http-log-source-in-panther) in Panther.
* Do not select a schema during HTTP source setup.
2. While viewing your log source's **Overview** tab, scroll down to the **Attach a schema to start classifying data** section.\
3. Choose from the following options:
* **I want to add an existing schema:** Choose this option if you already created a schema. Click **Start** in the tile.
* You will be navigated to the HTTP source edit page, where you can make a selection in the **Schemas - Optional** field:
HTTP source edit page
* **I want to generate a schema:** Select this option to generate a schema from live data. Click **Start** in the tile.
* Note that you may need to wait a few minutes after `POST`ing the events to the HTTP endpoint for them to be visible in Panther.
* On the page you are directed to, under **Raw Events**, you can view the raw data Panther has received within the last week:
HTTP Raw events
* This data is displayed from `data-archiver`, a Panther-managed S3 bucket that retains raw HTTP source logs for 15 days.
### **Infer a schema from raw data**
If you choose **I want to generate a schema** in the previous section, now you can infer a schema.
1. Once you see data populating within **Raw Events**, click **Infer Schema**.\
2. On the **Infer New Schema** modal that pops up, enter the:
* **New Schema Name:** Enter a descriptive name. It will always start with `Custom.` and must have a capital letter after.
3. Click **Infer Schema**.
* At the top of the page, you will see **'\' was successfully inferred**.
4. Click **Done**.\
* The schema will be placed in **Draft** mode until you're ready to publish it, after testing.
5. Click the draft schema's name to review its inferred fields.\
* Since the schema is in **Draft**, you can add, remove, and otherwise change fields as needed.\
### **Test the schema with raw data**
Once your schema is defined, you can proceed to test the schema configuration against raw data.
1. In the **Test Schemas** section at the top of the screen, click **Run Test**. \
2. In the **Test Schemas** pop-up modal, select the **Time Period** you would like to test your schema against, then click **Start Test**.\
* Depending on the time range and amount of data, the test may take a few minutes to complete.\
* Once the test is started, the results appear with the amount of matched and unmatched events.
* **Matched Events** represent the number of events that would successfully classify against the schema configuration.
* **Unmatched Events** represent the number of events that would not classify against the schema.
3. If there are **Unmatched Events**, inspect the errors and the JSON to decipher what caused the failures.\
* Click **Back to Schemas**, make changes as needed, and test the schema again.
4. Click **Back to Schemas**.
5. In the upper right corner, click **Save**.\
* The inferred schema is now attached to your log source.
* Log events that were sent to the HTTP source before it had a schema attached, which were used to infer the schema, are then ingested into Panther.
{% endtab %}
{% tab title="Manually" %}
## Creating a custom schema manually
To create a custom schema manually:
1. In the Panther Console, navigate to **Configure > Schemas.**
2. Click **Create New** in the upper right corner.
3. Enter a **Schema ID**, **Description**, and **Reference URL**.
* The Description is meant for content about the table, while the Reference URL can be used to link to internal resources.
4. Optionally enable **Automatic Field Discovery** by clicking its toggle `ON`. Learn more in [Enabling field discovery](#enabling-field-discovery).
5. In the YAML code block, write or paste your YAML log schema definition.
6. Click **Validate Syntax** at the bottom to verify your schema contains no errors.
* Note that syntax validation only checks the syntax of the Log Schema. It can still fail to save due to name conflicts.
7. Click **Save**.
You can now navigate to **Configure > Log Sources** and add a new source or modify an existing one to use the new `Custom.SampleAPI` \_Log Type. Once Panther receives events from this Source, it will process the logs and store the Log Events to the `custom_sampleapi` table.
You can also now write Rules to match against these logs and query them using the [Data Explorer](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/search/data-explorer.md).
{% endtab %}
{% endtabs %}
## Writing schemas
See the tabs below for instructions on writing schemas for JSON logs and for text logs.
Note that you can use the [`pantherlog` CLI tool](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/panther-developer-workflows/pantherlog.md#generating-a-schema-from-json-samples)[ ](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/panther-developer-workflows/pantherlog.md)to generate your Log Schema.
{% tabs %}
{% tab title="JSON Logs" %}
### Writing a schema for JSON logs
To parse log files where each line is JSON you have to define a log schema that describes the structure of each log entry.
You can edit the YAML specifications directly in the Panther Console or they can be [prepared offline in your editor/IDE of choice](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/reference.md#using-json-schema-in-an-ide). For more information on the structure and fields in a *Log Schema*, see the [Log Schema Reference](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/reference.md).
In the example schemas below, the first tab displays the JSON log structure and the second tab shows the Log Schema.
{% hint style="info" %}
**Note**: Please leverage the **Minified JSON Log Example** when using the `pantherlog` tool or generating a schema within the Panther Console.
{% endhint %}
{% tabs %}
{% tab title="JSON Log Example" %}
```json
{
"method": "GET",
"path": "/-/metrics",
"format": "html",
"controller": "MetricsController",
"action": "index",
"status": 200,
"params": [],
"remote_ip": "1.1.1.1",
"user_id": null,
"username": null,
"ua": null,
"queue_duration_s": null,
"correlation_id": "c01ce2c1-d9e3-4e69-bfa3-b27e50af0268",
"cpu_s": 0.05,
"db_duration_s": 0,
"view_duration_s": 0.00039,
"duration_s": 0.0459,
"tag": "test",
"time": "2019-11-14T13:12:46.156Z"
}
```
**Minified JSON log example**:
`{"method":"GET","path":"/-/metrics","format":"html","controller":"MetricsController","action":"index","status":200,"params":[],"remote_ip":"1.1.1.1","user_id":null,"username":null,"ua":null,"queue_duration_s":null,"correlation_id":"c01ce2c1-d9e3-4e69-bfa3-b27e50af0268","cpu_s":0.05,"db_duration_s":0,"view_duration_s":0.00039,"duration_s":0.0459,"tag":"test","time":"2019-11-14T13:12:46.156Z"}`
{% endtab %}
{% tab title="Log Schema Example" %}
```yaml
version: 0
fields:
- name: time
description: Event timestamp
required: true
type: timestamp
timeFormats:
- rfc3339
isEventTime: true
- name: method
description: The HTTP method used for the request
type: string
- name: path
description: The path used for the request
type: string
- name: remote_ip
description: The remote IP address the request was made from
type: string
indicators: [ ip ] # the value will be appended to `p_any_ip_addresses` if it's a valid ip address
- name: duration_s
description: The number of seconds the request took to complete
type: float
- name: format
description: Response format
type: string
- name: user_id
description: The id of the user that made the request
type: string
- name: params
type: array
element:
type: object
fields:
- name: key
description: The name of a Query parameter
type: string
- name: value
description: The value of a Query parameter
type: string
- name: tag
description: Tag for the request
type: string
- name: ua
description: UserAgent header
type: string
```
{% endtab %}
{% endtabs %}
{% endtab %}
{% tab title="Text logs" %}
### Writing a schema for text logs
Panther handles logs that are not structured as JSON by using a 'parser' that translates each log line into key/value pairs and feeds it as JSON to the rest of the pipeline. You can define a text parser using the `parser` field of the *Log Schema*. Panther provides the following parsers for non-JSON formatted logs:
| Name | Description |
| -------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- |
| [fastmatch](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/example-fastmatch.md) | Match each line of text against one or more simple patterns |
| [regex](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/example-regex.md) | Use regular expression patterns to handle more complex matching such as conditional fields, case-insensitive matching etc |
| [csv](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/example-csv.md) | Treat log files as CSV mapping colunm names to field names |
{% endtab %}
{% endtabs %}
## Managing custom schemas
### Editing a custom schema
Panther allows custom schemas to be edited. Specifically, you can perform the following actions:
* Add new fields.
* Rename or delete existing fields.
* Edit, add, or remove all properties of existing fields.
* Modify the `parser` configuration to fix bugs or add new patterns.
* [Archive or unarchive the schema](#archiving-and-unarchiving-a-custom-schema).
* [Enable or disable field discovery](#enabling-field-discovery).
{% hint style="info" %}
Note: After editing a field's `type`, any newly ingested data will match the new type while any previously ingested data will retain its type.
{% endhint %}
To edit a custom schema:
1. Navigate to your custom schema's details page in the Panther Console.
2. Click **Edit** in the upper right corner of the details page.
\*
```
```
3. Modify the YAML.
* Click **Diff View** in the upper right corner of the text editor to see the additions, edits, and subtractions via the code editor. It also includes the ability to copy or revert deleted lines.\
4. Click **Update** to submit your change.
Click **Validate Syntax** to check the YAML for structural compliance. Note that the rules will only be checked after you click **Update**. The update will be rejected if the rules are not followed.
#### Update related detections and saved queries
Editing schema fields might require updates to related detections and saved queries. Click on the related entities in the alert banner displayed above the schema editor to view, update, and test the list of affected detections and saved queries.
#### **Query implications**
Queries will work across changes to a **Type** provided the query does not use a function or operator which requires a field type that is not castable across **Types**.
* **Good example**: The **Type** is edited from `string` to `int` where all existing values are numeric (i.e. `"1"`). A query using the function `sum` aggregates old and new values together.
* **Bad example**: The **Type** is edited from `string` to `int` where some of the existing values are non-numeric (i.e. `"apples"`). A query using the function `sum` excludes values that are non-numeric.
#### Query castability table
This table shows which **Types** can be cast as each **Type** when running a query. Schema editing allows any **Type** to be changed to another **Type**.
Type From -> To
boolean
string
int
bigint
float
timestamp
boolean
same
yes
yes
yes
no
no
string
yes
same
numbers only
numbers only
numbers only
numbers only
int
yes
yes
same
yes
yes
numbers only
bigint
yes
yes
yes
same
yes
numbers only
float
yes
yes
yes
yes
same
numbers only
timestamp
no
yes
no
no
no
same
### Archiving and unarchiving a custom schema
You can archive and unarchive custom schemas in Panther. You might choose to archive a schema if it's no longer used to ingest data, and you do not want it to appear as an option in various dropdown selectors throughout Panther. In order to archive a schema, it must not be in use by any log sources. Schemas that have been archived still exist indefinitely; it is not possible to permanently delete a schema.
Archiving a schema does not affect any data ingested using that schema already stored in the data lake—it is still queryable using [Data Explorer](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/search/data-explorer.md) and [Search](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/search/search-tool.md). By default, archived schemas are not shown in the schema list view (visible on **Configure** > **Schemas**), but can be shown by modifying **Status**, within **Filters**, in the upper right corner. In [Data Explorer](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/search/data-explorer.md), tables of archived schemas are not shown under **Tables**.
Attempting to create a new schema with the same name as an archived schema will result in a name conflict, and prompt you to instead unarchive and edit the existing schema.
To archive or unarchive a custom schema:
1. In the Panther Console, navigate to **Configure** > **Schemas**.
* Locate the schema you'd like to archive or unarchive.
2. Click the three dots icon in the upper right corner of the tile, and select **Archive** or **Unarchive**.
* If you are archiving a schema and it is currently associated to one or more log sources, the confirmation modal will prompt you to first detach the schema. Once you have done so, click **Refresh**.\
3. On the confirmation modal, click **Continue**.
### Testing a custom schema
{% hint style="info" %}
The "Test Schema against sample logs" feature found on the Schema Edit page in the Panther Console supports Lines, CSV (with or without headers), JSON, JSON Array, and CloudWatch Logs. See [Stream Types](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/reference.md#stream-type) for examples.
Additionally, the above log formats can be compressed using the following formats:
* gzip
* zstd (without dictionary)
Multi-line logs are supported for JSON and JSONArray formats.
{% endhint %}
Need to validate that a custom schema will work against your logs? You can test sample logs by following this process:
1. In the Panther Console, go to **Configure > Schemas**.
2. Click on a custom schema**.**
3. In the schema details page, scroll to the bottom of the page where you'll be able to upload logs.
### Enabling field discovery
Log source schemas in Panther define the log event fields that will be stored in Panther. When field discovery is enabled, data from fields in incoming log events that are *not* defined in the corresponding schema will not be dropped—instead, the fields will be identified, and the data will be stored. This means you can subsequently query data from these fields, and write detections referencing them.
Field discovery can only be enabled for [JSON ](#writing-a-schema-for-json-logs)and [CSV with header](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types/example-csv.md#csv-logs-with-header) log schemas. Field discovery is currently only available for custom schemas, not Panther-managed ones. See [additional limitations of field discovery below](#limitations).
#### Handling of special characters in field names
If a field name contains a special character—a character that is not alphanumeric, an underscore (`_`) , or a dash ( `-`)—it will be transliterated using the algorithm below:
* `@` to `at_sign`
* `,` to `comma`
* `` ` `` to `backtick`
* `'`to `apostrophe`
* `$` to `dollar_sign`
* `*` to `asterisk`
* `&` to `ambersand`
* `!` to `exclamation`
* `%` to `percent`
* `+` to `plus`
* `/` to `slash`
* `\` to `backslash`
* `#` to `hash`
* `~` to `tilde`
* `=` to `eq`
All other ASCII characters (including `space`) will be replaced with an underscore (`_`). Non-ASCII characters are transliterated to their closest ASCII equivalent.
This transliteration affects only field names; values are not modified.
#### Limitations
Field discovery currently has the following limitations:
* The maximum number of top-level fields that can be discovered is 1,000. Within each `object` field, a maximum of 1,000 fields can be discovered.
* There is no limitation on the number of overall fields discovered.
## Uploading log schemas with the Panther Analysis Tool
If you choose to maintain your log schemas outside of Panther, for example in order to keep them under version control and review changes before updating, you can upload the YAML files programmatically with the [Panther Analysis Tool](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/panther-developer-workflows/ci-cd/deployment-workflows/pat.md).
The uploader command receives a base path as an argument and then proceeds to recursively discover all files with extensions `.yml` and `.yaml`.
{% hint style="warning" %}
It is recommended to keep schema files separately from other unrelated files, otherwise you may notice several unrelated errors for attempting to upload invalid schema files.
{% endhint %}
```
panther_analysis_tool update-custom-schemas --path ./schemas
```
The uploader will check if an existing schema exists and proceed with the update or create a new one if no matching schema name is found.
{% hint style="danger" %}
The `schema`field must always be defined in the YAML file and be consistent with the existing schema name for an update to succeed. For a list of all available CI/CD fields see our [Log Schema Reference](https://docs.panther.com/data-onboarding/custom-log-types/reference#ci-cd-schema-fields).
{% endhint %}
{% hint style="info" %}
The uploaded files are validated with the same criteria as Web UI updates.
{% endhint %}
## Troubleshooting Custom Logs
Visit the Panther Knowledge Base to [view articles about custom log sources](https://help.panther.com/Data_Sources/Custom_Logs) that answer frequently asked questions and help you resolve common errors and issues.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.panther.com/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/custom-log-types.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
* All unsuccessfully matched logs will appear under **Unmatched;** each log will display the error message and the raw log.
11. Click **Save** to publish the schema.
{% hint style="info" %}
Panther will infer from all logs uploaded, but will only display up to 100 logs to ensure fast response time when generating a schema.
{% endhint %}
{% endtab %}
{% tab title="S3 data received in Panther" %}
## Inferring a custom schema from S3 data received in Panther
You can generate and publish a schema for a custom log source from live data streaming from an S3 bucket into Panther. You will first [view your S3 data](#view-raw-s3-data) in Panther, then [infer a schema](#infer-a-schema-from-raw-data), then [test the schema](#test-the-schema-with-raw-data).
### **View raw S3 data**
After onboarding your S3 bucket into Panther, you can view raw data coming into Panther and infer a schema from it:
1. Follow the instructions to [onboard an S3 bucket onto Panther](/~/changes/Dd8nx2iqd1Pp2OzWJaWk/data-onboarding/data-transports/aws/s3.md) without having a schema in place.
2. While viewing your log source's **Overview** tab, scroll down to the **Attach a schema to start classifying data** section.\
3. Choose from the following options:
* **I want to add an existing schema:** Choose this option if you already created a schema and you know the S3 prefix you want Panther to read logs from. Click **Start** in the tile.
* You will see a **S3 Prefixes & Schemas** popup modal:\
* **I want to generate a schema from raw events:** Select this option to generate a schema from live data in this bucket and define which prefixes you want Panther to read logs from. Click **Start** in the tile.
* Note that you may need to wait up to 15 minutes for data to start streaming into Panther.
* On the page you are directed to, you can view the raw data Panther has received at the bottom of the screen:\
* This data is displayed from `data-archiver`, a Panther-managed S3 bucket that retains raw logs for up to 15 days for every S3 log source.
* Only raw log events that were placed in the S3 bucket *after* you configured the source in Panther will be visible, even if you've set the timespan to look further back.
* If your raw events are JSON-formatted, you can view them as JSON by clicking **View JSON** in the left-hand column.
### **Infer a schema from raw data**
If you chose to **I want to generate a schema from raw events** in the previous section, now you can infer a schema.
1. Once you see data populating in **Raw Events,** you can filter the events you'd like to infer a schema from by using the string Search, S3 Prefix, Excluded Prefix, and/or Time Period filters at the top of the **Raw Events** section.
2. Click **Infer Schema** to generate a schema.\
3. On the **Infer New Schema** modal that pops up, enter the following:
* **New Schema Name:** The name of the schema that will map to the table in the data lake once the schema is published.
* The name will always start with `Custom.` and must have a capital letter after.
* **S3 Prefix:** Use an existing prefix that was set up prior to inferring the schema or a new prefix.
* The prefix you choose will filter data from the corresponding prefix in the S3 bucket to the schema you've inferred.
* If you don't need to specify a specific prefix, you can leave this field empty to use the catch-all prefix that is called `*`.\
4. Click **Infer Schema**.
* At the top of the page, you will see **'\
