The Open-AudIT API

Introduction

Open-AudIT has a JSON Restful API to be used both in the web interface and via JSON requests.

Open-AudIT's API

Open-AudIT's API is base upon http://jsonapi.org with the intention of providing simple and intuitive access in a manner familiar to developers.

In addition to this API, the web interface will use the same request format and supply some additional actions (eg: HTML forms for creating items).

 

Access Model

The API uses a cookie. You can request a cookie by sending a POST to the URL below, containing the username and password attributes and values:

http://{server}/open-audit/index.php/logon

POSTing data

To create a resource, you should POST the required data.

When POSTing data, you must include an access token. An access token is generated with every request type, so make a GET (for example) and Accept: application/json, parse the response for meta→access_token, and include that with your request. This should be placed in the field data[access_token], IE, the top level.

The format of your data should be in the form:

data[attributes][ATTRIBUTE_NAME]

You should substitute the required column (eg, org_id) for ATTRIBUTE_NAME.

In the case where we store several fields (usually in JSON format) inside a BIGTEXT MySQL field (eg: credentials.credentials - the credentials column in the credentials table), you should use the format:

data[attributes][credentials][credentials][username]

Som examples are at the bottom of this page.

All endpoints also have a minimum list of required fields. These are:

 

applications: name,org_id

attributes: name,org_id,type,resource,value

baselines: name,org_id

buildings: name,org_id,location_id

clouds: name,org_id,type,credentials,options

clusters: name,org_id

collectors: name,org_id,type,host,community,username,password

connections: name,org_id

credentials: name,org_id,type,credentials

dashboards: name,org_id,options,sidebar

devices: name,org_id

discoveries: name,org_id,type

discovery_scan_options: name,org_id,ping,service_version,filtered,open|filtered,timing,nmap_tcp_ports,nmap_udp_ports

fields: name,org_id,type

files: name,org_id,path

groups: name,org_id,sql

integrations: name,org_id,attributes,fields

ldap_servers: name,org_id,lang,host,port,secure,domain,type,version,use_auth,use_roles,refresh

licenses: name,org_id,org_descendants,purchase_count,match_string

locations: name,org_id

networks: name,org_id,network

orgs: name,parent_id

queries: name,org_id,sql,menu_category,menu_display

racks: name,org_id,ru_height

rack_devices: rack_id,device_id,position,height

roles: name,permissions

rules: name,org_id

scripts: name,org_id,options,based_on

summaries: name,org_id,table,column,menu_category

tasks: name,org_id,type,sub_resource_id,uuid,enabled,minute,hour,day_of_month,month,day_of_week

users: name,org_id,lang,roles,orgs

widgets: name,org_id,type

 

An example JSON POST body is below. This should be attached to the "data" form item.

{ "access_token": "bbc0c85653fdc4b83d108cba7641bfcbbc77586dfb8f32d08973770a90fe", "type": "discoveries", "attributes": { "name": "My Test Discovery", "type": "subnet", "subnet": "192.169.1.150" "org_id": 1, "scan_options": {<removed for brevity>}, "match_options": {<removed for brevity>}, } } }

 

The Endpoints

At present, we have endpoints for nearly every collection. They are listed here - Collections.

Options

Format

Using the format option is useful when using a web browser but you wish to see the result in JSON format. Adding format=json achieves this. If you only want the actual data in JSON, format=json_data will do the trick. Normally a web browser will set its accept header to HTML, so in that case, we return the rendered page. Using an API to retrieve JSON you should set the accept header to "json/application". You can override this by providing the format option in the URL.

We tend to use the Google Chrome extension called Postman for testing actual restful queries. You might like to install and test with that. http://www.getpostman.com.

Action

NOTE - Removed from 5.0.0.

When using the API the default action is determined according to the format and URL. You can override this by providing the 'action' option in the URL. An example of this is when creating a new item. You would normally use POST to /item but in the case of a web user, you need a web form to be able to fill out the item details. In that case, there is no facility for this in a typical JSON Restful API. We work around this by providing action=create in a GET request for the URL. IE - http://{server}/omk/open-audit/networks?action=create. The default action if nothing matches below is to return a collection of items.

API Routes

Request Method

ID

Action

Resulting Function

Permission Required

URL Example

Notes

Request Method

ID

Action

Resulting Function

Permission Required

URL Example

Notes

POST

n

 

create

{collection}::create

/{collection}

Insert a new {collection} entry.

GET

y

 

read

{collection}::read

/{collection}/{id}

Returns a {collection} details.

PATCH

y

 

update

{collection}::update

/{collection}/{id}

Update an attribute of a {collection} entry.

DELETE

y

 

delete

{collection}::delete

/{collection}/{id}

Delete a {collection} entry.

GET

n

 

collection

{collection}::read

/{collection}

Returns a list of {collection}.

Web Application Routes

Request Method

ID

Action

Resulting Function

Permission Required

URL Example

Notes

Request Method

ID

Action

Resulting Function

Permission Required

URL Example

Notes

GET

n

create

create_form

{collection}::create

/{collection}/create

Displays a standard web form for submission to POST /{collection}.

GET

n

import

import_form

{collection}::create

/{collection}/import

Displays a standard web form for submission to POST /{collection}/import.

POST

n

import

import

{collection}::create

/{collection}/import

Import multiple {collection} using a CSV.

GET

y

execute

execute

(collection)::see below

/{collection}/{id}/execute

Some collections can be executed - queries, etc - see below.

 

Execute permissions required per endpoint

Endpoint

Permission

Endpoint

Permission

baselines

read

clouds

read

dashboards

read

database

update

discovery

update

groups

read

queries

read

summaries

read

tasks

read

 

Sort

To sort by a database column, use "sort={attribute}". To reverse sort, insert a minus, thus "sort=-{attribute}".

sort=[-]{attribute}

Current

NOTE - Removed from 5.0.0. Please use components.current=n or components.current=IN('y','n') instead (if required).

By default, only attributes with "current=y" are retrieved. To override this, set current as below.

GroupBy

NOTE - Removed from 5.0.0.

Limit

When requesting JSON, by default no limit is set.

When requesting screen display, the limit is set to 1000 by default.

Offset

The offset is the count of devices you wish to return data from.

Properties

The requested properties should be in a comma-separated list. Properties should be fully qualified - ie, system.hostname (not just hostname).

NOTE - From 5.0.0 onwards, the system table has been replace by the devices table - so devices.name, not system.name.

You can also specify properties using the below format.

 

Filter

To filter by a property value, use the property name. Operators that should precede the value are !=, >, >=, <, <=, 'like' and '!like'. If no operator is specified, the default is =. Properties should be fully qualified - ie, system.hostname (not just hostname).

End Points

All endpoints URLs for prior to v5 are of the format http://{server}/omk/open-audit/{endpoint}

NOTE - From 5.0.0 all endpoint URLs are of the form - http://{server}/open-audit/index.php/{endpoint}

Devices

NOTE - From 5.0.0 the sub_resource item has been replaced by the components endpoint.

Type

Endpoint v4

v5

 

 

Type

Endpoint v4

v5

 

 

GET

/system

/devices

Return a collection of devices with the default set of columns from the system table (system.system_id, system.icon, system.man_type, system.hostname, system.domain, system.man_ip_address, system.man_description, system.man_os_family, system.man_status)

 

GET

/system/{id}

/devices/{id}

Return an individual devices details.

 

GET

/system?sub_resource={sub_resource name}


/components?components.type={sub_resource name}

To return all items in a sub_resource for a collection of devices. If you wanted all software you would use http://{server}/open-audit/index.php/devices?sub_resource=software

 

GET

/system/{id}?sub_resource={sub_resource name}


/components?components.type={sub_resource name}&components.device_id={id}

To return all items in a sub_resource for a specific device.

 

GET

/system/{id}?sub_resource={sub_resource name}&sub_resource_id={sub_resource id}


/components/{sub_resource id}?components.type={sub_resource name}

To return a specific sub_resource item.

 

Device sub_resource names / component types

 

NAME

NAME

NAME

NAME

NAME

NAME

audit_log
bios
change_log
credential
disk
dns
edit_log
ip
log
memory
module
monitor
motherboard

netstat
network
optical
pagefile
partition
print_queue
processor
radio
route
san
scsi
server
server_item 

service
share
software
software_key
sound
task
user
user_group
variable
video
vm
windows

Examples

NOTE - Below are v5.0.0 and onward URLs.

NOTE #3 - You should substitute items in the URL enclosed in {} brackets with the relevant items for your environment (and do not include the actual brackets themselves!).

Retrieve all devices with the standard columns:

Retrieve all devices running Windows.

Retrieve the first 10 devices running Windows ordered by hostname

Retrieve the properties id, ip, hostname, domain, type from all devices

Retrieve all details about the device with id 88.

Retrieve a list of devices in the 192.168.1.0/24 subnet

Retrieve a list of devices with OS Name like Windows 2008

 

CURL Examples

Logging in

Creating Credentials

Retrieving a List of Credentials

Update attributes

NOTE - The curly brackets in the data filed should be used as-is (not replaced as per other examples above).