Nullgap.dev Docs

Help

The support channel

Contact Nullgap support from the console, follow your requests, and let support see your workspace for a limited time, only with your consent.

When something is unclear, broken or looks wrong, you can reach the Nullgap support team from inside the console. Requests are written, followed and answered there. If seeing your workspace would help, an admin can let support look at it for a limited time, reading only or with a few support actions that remove nothing; support has no such access otherwise.

Before you write, getting help shows where to look first: a run's log, a skip reason, a webhook's deliveries.

#Send a request

  1. Open Help at the bottom of the menu. To attach a finding, a review or a pentest, open it first and then choose Help: the console remembers the page you came from.
  2. On the Contact support tab, write a subject (up to 150 characters) and a message (up to 5,000).
  3. Choose what it is about:

    Choicecategory
    A question about how to do somethinghow_to
    Something is not workingnot_working
    A finding looks wrongfinding_wrong
    Billing or planbilling
    Security concernsecurity
    Otherother
  1. Choose how urgent it is:

    ChoiceurgencyUse it when
    NormalnormalMost requests
    Blocking my workblockingYou cannot continue until it is answered
    Production security incidentincidentA security incident in production is happening now. Use it for nothing else
  1. Check the box Sent along with your message. The console adds the page you were on, the finding, review or pentest you had open, and the reference of the last request that failed, so that support can find what you mean. Untick anything you would rather not send.
  2. Choose Send to support.

Do not paste passwords, tokens or keys into a request. Messages are plain text: nothing you type is treated as formatting, and files cannot be attached.

If e-mail is set up for your installation, you get a short confirmation with the request's id.

#Follow your requests

The My requests tab lists your requests with their status.

StatusstatusMeaning
OpenopenSent. Nobody at Nullgap has answered yet
Waiting on youwaiting_customerSupport has answered and is waiting for your reply
Waiting on Nullgapwaiting_supportYou have replied. The next step is Nullgap's
ResolvedresolvedClosed. You can reopen it, or simply reply to it

Open a request to read the thread, reply, Mark resolved or Reopen. A reply to a resolved request reopens it.

Who sees what:

  • A member sees the requests they opened.
  • Admins and owners see every request of the organisation.
  • Replies from Nullgap are signed "Nullgap support". The name of the person who answered is not shown.

#Replies arrive in the console

Answers are read and written in the console. If e-mail is set up for your installation, you get a short notice when support replies, with a link back to the request. The notice never contains the reply itself, the subject, or anything about your findings, and answering that e-mail reaches nobody: reply in the console.

#Letting support see your workspace

Support cannot open your findings, reviews or pentests. When you attach a finding to a request, support sees its id and nothing more.

If seeing them would help, an admin or owner can allow it. This is your organisation's decision, it is limited in time, and it can be ended at any moment.

  1. Open Help, Support access. Or open the request itself, which has the same control; you can also tick it while writing a new request.
  2. Choose how long: 1 hour, 24 hours or 7 days.
  3. Choose whether to Include support actions (below). Without them, support only reads.
  4. Choose Allow access. Because this opens your workspace to someone outside your organisation, you are asked for a fresh second-factor check first. See the step-up check.

Access given from a request's page is tied to that request and ends when the request is resolved, by either side, even if the time has not run out. Access given from the Support access tab is not tied to a request and ends when its time is up.

One grant is active at a time. A new one replaces the one before it.

#When support asks for access

Support can also ask. A request for access says why, for how long, and whether support actions are wanted. Every owner and admin sees it under Help, Support access (and gets an e-mail if e-mail is set up). One of them chooses:

  • Approve: you choose the duration and whether support actions are included, whatever the request asked for, and confirm with a fresh second-factor check. The approval is your own grant, exactly as if you had given it yourself.
  • Decline: nothing is opened.

A request nobody answers lapses after 72 hours. Support can have one request waiting at a time, cannot ask again for 24 hours after a decline, and can ask at most 3 times in 7 days. Requests are logged as support.access_requested, support.access_request_approved, support.access_request_declined and support.access_request_withdrawn.

#What support can read while it is on

Support can readSupport can never see
Findings: the list, and one finding with its evidence and code excerptStored credentials: code-host tokens, integration secrets, API keys, webhook signing secrets
Pull request reviews and pentests: the lists, one run with its findings, and their logsThe address of a webhook or an inbound hook (only its host is shown), or a report's share link
Tool runs and skill runs, fix runs with their reasons, and the job queueAn invitation's link
Review settings and the list of repositoriesSingle sign-on and other sign-in settings
People: members with their role and last activity, and pending invitationsYour repository. Support does not have your code
Connections, integrations and outbound webhooks, with their health

What support sees are the same forms of each record that your own console shows, with stored secrets left out.

#Support actions

Reading changes nothing. If you Include support actions, support may also do four things an admin could do themselves, none of which removes anything:

  • start a failed job again;
  • cancel a run that is stuck;
  • refresh a connection's list of repositories;
  • send an invitation again.

Each action is checked against the live grant at the moment it is done, and each one is written to your audit log as support.action. Support can do nothing else in your workspace: it cannot change settings, findings, people or billing, and it cannot delete anything.

#How you can tell

  • While access is on, every admin sees a banner on every page saying what support may do (reading only, or reading and the support actions), until when, and who allowed it. The banner links to Manage or end it.
  • Every page support opens is written to your audit log as support.access_used by "Nullgap support", with what was opened. Every support action is written as support.action.
  • Giving and ending access are logged as support.access_granted and support.access_revoked, with who did it.

#Ending it

Choose End access now on the Support access tab, which the banner links to, and confirm. It takes effect at once. The tab also lists the ten most recent earlier grants, with who gave each one and when it ended.

#Limits

LimitValue
New requests5 per person and 20 per organisation, per hour
Replies30 per person and 100 per organisation, per hour
Subject150 characters
Message5,000 characters
Messages in one request200. After that, open a new request and mention the old one
AttachmentsNone

Above a rate limit the console says to wait a while and try again.

#API keys cannot use the support channel

A support request is a conversation between people, so it needs a signed-in person. Listing, opening, sending, replying, resolving and reopening requests all refuse an API key with 403 forbidden, and so does giving support access. There is no MCP tool for any of it.

Two calls about support access do work with a key that has the admin scope: reading whether access is on, and ending it.

curl -s -H "Authorization: Bearer $NULLGAP_API_KEY" \
  "$NULLGAP_URL/api/orgs/$NULLGAP_ORG/support/access"

The answer has active (the current grant, or null) and history. DELETE /api/orgs/:org/support/access/:id ends the grant with that id.

One address needs no sign-in at all, because error pages use it. It says how support can be reached on your installation:

curl -s "$NULLGAP_URL/api/support/info"
{
  "email": "support@nullgap.example",
  "statusUrl": "https://status.nullgap.example",
  "docsUrl": "https://nullgap.example/docs",
  "emailNotices": true
}
FieldMeaning
emailThe support address, or null when requests are handled in the console only
statusUrl, docsUrlLinks to the status page and the documentation, or null
emailNoticesWhether the installation sends e-mail notices at all

The same address and links are shown on the Help page and on error pages.

#If your workspace is suspended

You can still open Help, send and answer requests, resolve them, and give or end support access. Everything else in a suspended workspace is read-only.

#Your data

  • Support requests and their threads belong to your organisation. They are included when an owner exports the organisation's data, together with the record of support-access grants, and they are removed when an owner erases all data. See data retention.
  • The retention period you set for findings does not delete support requests.
  • Your organisation's audit log records what its members did: support.request_opened, support.request_resolved and support.request_reopened.

#Reporting a security issue in Nullgap

Use the category Security concern, and say at the start that it is a report about Nullgap itself. What to include, and what to leave out, is under reporting a security issue in Nullgap.

Nullgap documentation