Administration
Retention, export and erase
Choose how long Nullgap keeps your findings and run history, take a copy of everything it stores, and erase it.
This page covers the controls under Settings, Data protection: how long data is kept, the framework and contacts you can record, and the owner's tools to export and erase.
For what Nullgap stores in the first place, and what it never stores, see how your code is handled.
#Who can do what
| Action | Who | With an API key |
|---|---|---|
| See the settings and the handling overview | Every member | Yes, any scope |
| Change retention, framework and contacts | Admin | Yes, admin scope |
| Export everything | Owner | No, console only |
| Erase data | Owner, with a fresh second-factor check | No, console only |
Members see the page read-only.
#Retention
By default, findings and run history are kept until you delete them. Setting a retention period makes Nullgap delete what is older than that, automatically.
| Setting | Value |
|---|---|
| Field | dataProtection.retentionDays |
| Allowed | A whole number of days from 7 to 3650, or null to keep until deleted by hand |
| Default | null |
| Console | Retention: Keep until deleted by hand, or Delete after 30, 90, 180 or 365 days, or a custom number |
#What the automatic purge deletes
Each run of the purge deletes, for an organisation with a retention period:
| Deleted | When |
|---|---|
| Pull request reviews and their logs | Created before the cut-off, and not queued or running |
| Pentests, their logs and their reports | Created before the cut-off, and not queued or running |
| Resolved findings (patched, accepted risk, false positive) | Resolved before the cut-off |
| The remembered pattern of a dismissed finding | Together with that finding |
| Findings that were held back by an earlier dismissal | Not seen since the cut-off |
| Fix runs | Created before the cut-off, and not queued or running |
| Scanning tool runs and their logs | Created before the cut-off, and not queued or running |
| Any other run log | Created before the cut-off |
The cut-off is the current time minus the retention period.
What the purge never deletes:
- Open findings. They are still in use, however old they are. They become eligible only once resolved, counted from the day they were resolved.
- The audit log.
- Repositories, applications and threat models, custom checks, automations, schedules, integrations, people, seats, the billing ledger and settings.
A deleted finding's dismissal memory goes with it, so the same flaw can be reported again by a later run. See dismissal memory.
#How often it runs
- When the service starts, and then every six hours.
- Immediately when you save the data protection settings, so a shorter period takes effect at once.
When a run deletes something, the audit log records data.retention with the actor nullgap and a count per kind. The time of the last run is shown on the page and returned as lastPurgeAt.
Deletion is permanent. Export first if you need a copy.
#Set it through the API
curl -s -X PATCH -H "Authorization: Bearer $NULLGAP_API_KEY" \
-H "Content-Type: application/json" \
-d '{"dataProtection":{"retentionDays":180}}' \
"$NULLGAP_URL/api/orgs/$NULLGAP_ORG/settings"
The answer is the organisation's settings after the change. Send "retentionDays": null to go back to keeping data until it is deleted by hand. See the organisation reference.
#Framework selection
Under Framework you can record the privacy law or standard your organisation works under (dataProtection.framework). The list covers privacy laws, AI governance frameworks, security standards and sector regimes; none is the default.
What selecting a framework changes:
- Retention gets a starting value. If no retention period is set and you do not send one in the same request, it is set to 365 days. You can change it afterwards, including back to
null. - The overview shows references. For the frameworks with a published mapping, each entry in the handling overview lists the provisions it relates to. For the others, the entries are listed without references.
- The overview gains one entry about a policy pack for that framework on the AI processing layer, with the status the service reports for it.
What it does not change: the protections described under security and privacy apply whichever framework is selected, and with none.
#The handling overview
GET /api/orgs/:org/compliance returns what the Data protection page shows:
curl -s -H "Authorization: Bearer $NULLGAP_API_KEY" \
"$NULLGAP_URL/api/orgs/$NULLGAP_ORG/compliance"
Abridged:
{
"framework": "gdpr",
"clauseMapped": true,
"retentionDays": 180,
"lastPurgeAt": "2026-11-16T06:00:00.000Z",
"zeroRetentionCoverage": 100,
"controls": [
{
"id": "retention",
"title": "Automatic deletion after a set period",
"status": "configurable",
"evidence": "Deleting after 180 days. Last run 2026-11-16.",
"references": ["Art. 5(1)(e) storage limitation"]
}
]
}
Each entry in controls has a status:
| Status | Meaning |
|---|---|
enforced | Done by the product; you cannot switch it off |
configurable | Built in, and set by your organisation's own settings |
contract | Covered by an agreement or a process rather than by code |
customer | Your organisation's own obligation; Nullgap provides what you need to meet it |
planned | Not available yet |
evidence says in plain words what backs the status for your organisation right now. available (not shown) lists every framework that can be selected. The full shape is in the data protection reference.
#Contacts
Under Contacts you can record two addresses:
| Console field | API field | Used for |
|---|---|---|
| Privacy contact | dataProtection.privacyContactEmail | Who data-subject requests and privacy questions go to at your organisation |
| Security-incident contact | dataProtection.breachContactEmail | Who Nullgap notifies if a security incident affects your organisation's data |
Each must be an e-mail address, or empty. The product stores the two addresses and shows them in the handling overview; it does not e-mail them automatically. They tell Nullgap's staff whom to reach. Set the incident contact to a monitored mailbox rather than one person.
#Export everything
An owner can download everything Nullgap stores for the organisation: Your data, Export everything. The result is a single JSON file named after the organisation.
It contains:
- the organisation's id, address, name, plan and settings;
- repositories, applications and their threat models;
- reviews, pentests, reports (with the record each pentest's reports are produced from), fix runs and run logs;
- findings, the remembered patterns of dismissed findings, and held-back findings;
- automations and pentest schedules;
- the audit log;
- the billing ledger, members and seats;
- support requests with their threads, and the record of support-access grants;
- your own skills with the text of each, which skills are switched off and where, and on-demand skill runs with their output and the text you supplied to them;
- scanning tool runs with their leads and documents, with credentials masked.
Left out entirely: integration credentials, code-host connections, outbound webhooks, API keys, custom checks, the files uploaded for a skill comparison, and the single sign-on and SCIM set-up.
Each record is written in the form the API gives it, not as it is stored. Stored secrets are not in the file in any form, encrypted or not: a pentest that was given a test account for live testing says only that it has one. The link of a shared report is not in the file either; a report says whether it is shared. The usage and cost fields of runs are left out.
The export is recorded in the audit log as data.exported. The endpoint is GET /api/orgs/:org/data/export; it needs a signed-in owner, so it cannot be called with an API key.
For findings alone, in formats made for other tools, see exports.
#Erase data
An owner can erase the organisation's data: Your data, Erase data…. Erasing cannot be undone, and nothing removed can be recovered afterwards.
Erasing empties the workspace and keeps the workspace itself, its members and its audit log. To remove the workspace as well, see deleting a workspace.
- Choose what to erase (the scope).
- Type the organisation's address (its slug) to confirm.
- Choose Erase permanently. If your account has a second factor, you are asked to confirm with it first.
#Scopes
| Scope | Console label | Removes | Keeps |
|---|---|---|---|
findings | Findings and run history | Findings, dismissal memory, held-back findings, logs, pull request reviews, pentests, reports, fix runs, on-demand skill runs, scanning tool runs | Everything else |
all | Everything except members, billing and the audit log | Everything in findings, plus applications and threat models, automations, integrations, outbound webhooks and their delivery records, the Slack assistant and linked Slack users, pentest schedules, repositories, uploaded archives, code-host connections, support requests with their threads and support-access grants, and custom skills with their switches and the files uploaded for a skill comparison | See below |
Kept in both scopes:
- the organisation itself and its settings;
- members, seats and the billing ledger;
- the audit log;
- API keys, the single sign-on and SCIM set-up, and custom checks.
A review, pentest, fix run, skill run or scanning tool run that is queued or running is not erased. Cancel it or wait for it to finish, then erase again.
With the all scope, Nullgap also replaces the organisation's encryption salt after the records are gone. A copy of one of the deleted encrypted secrets that still existed somewhere could then no longer be opened. The few secrets that are kept, such as the single sign-on client secret, are re-encrypted in the same step and keep working. If a pentest with a live-testing test account is still queued or running, the salt is left as it is, and the audit entry says so. See encryption.
#The request
The console sends POST /api/orgs/:org/data/erase with:
{ "scope": "findings", "confirm": "acme" }
| Field | Value |
|---|---|
scope | findings or all |
confirm | The organisation's slug, exactly. Anything else is refused with 400 |
The answer counts what was removed, by kind:
{ "deleted": { "findings": 212, "logs": 96, "reviews": 88, "pentests": 4, "reports": 6 } }
The audit log records data.erased with the scope, the counts, and whether the encryption salt was replaced.
#Console-only actions
These need a signed-in owner and refuse every API key:
- exporting everything;
- erasing data, which also asks for a fresh second-factor check. See the step-up check.
Changing retention, the framework and the contacts can be done with an API key that has the admin scope, as shown above.
#Limits to know about
- Retention and erasure act on what Nullgap stores. They do not remove what Nullgap wrote elsewhere on your behalf: comments, checks and fix pull requests on your code host, issues in your tracker, messages in Slack, or reports you downloaded or shared.
- Retention and erasure remove records from the live service. They do not reach backups of the service made earlier. Ask Nullgap how long backups are kept if that matters to you.
- To close the organisation altogether, contact Nullgap: there is no control in the console that deletes the organisation and its members.
Nullgap documentation