akismet.md 1.85 KB
Newer Older
1 2
# Akismet

3
> *Note:* Before 8.11 only issues submitted via the API and for non-project
4 5
members were submitted to Akismet.

6
GitLab leverages [Akismet](http://akismet.com) to protect against spam. Currently
7 8
GitLab uses Akismet to prevent the creation of spam issues on public projects. Issues
created via the WebUI or the API can be submitted to Akismet for review.
9

10 11
Detected spam will be rejected, and an entry in the "Spam Log" section in the
Admin page will be created.
12

13 14 15 16 17 18 19 20 21 22 23 24
Privacy note: GitLab submits the user's IP and user agent to Akismet. Note that
adding a user to a project will disable the Akismet check and prevent this
from happening.

## Configuration

To use Akismet:

1. Go to the URL: https://akismet.com/account/

2. Sign-in or create a new account.

25
3. Click on **Show** to reveal the API key.
26 27 28

4. Go to Applications Settings on Admin Area (`admin/application_settings`)

29
5. Check the **Enable Akismet** checkbox
30 31 32 33 34 35

6. Fill in the API key from step 3.

7. Save the configuration.

![Screenshot of Akismet settings](img/akismet_settings.png)
36 37 38 39 40 41 42 43 44 45 46


## Training

> *Note:* Training the Akismet filter is only available in 8.11 and above.

As a way to better recognize between spam and ham, you can train the Akismet
filter whenever there is a false positive or false negative.

When an entry is recognized as spam, it is rejected and added to the Spam Logs. 
From here you can review if they are really spam. If one of them is not really
47
spam, you can use the **Submit as ham** button to tell Akismet that it falsely 
48 49 50 51 52
recognized an entry as spam.

![Screenshot of Spam Logs](img/spam_log.png)

If an entry that is actually spam was not recognized as such, you will be able
53
to also submit this to Akismet. The **Submit as spam** button will only appear
54 55 56 57 58
to admin users.

![Screenshot of Issue](img/submit_issue.png)

Training Akismet will help it to recognize spam more accurately in the future.