1
Start a new scan
Click New Scan in the sidebar and pick the repository, or open the
repository from Repositories and start a scan there.
2
Scan the default branch
Run a deep scan on the branch your team trusts as the
baseline (usually
main). Keep the first run simple: one repository, one
branch. You can add related repositories later for a multi-repo scan.3
4
Review results
Open the Findings tab to triage findings, the Endpoints tab for
the attack surface, and the Wiki tab for the codebase overview. See
Findings & remediation.
Run a scan from the API
Trigger the same scan from a script, a CI job, or a developer machine withPOST /scans. Use an
API key with the scans.run permission.
commitpins the scan to one commit, full or abbreviated SHA. Without it, Gecko scans the tip ofbranchat launch.branchlabels the scan. Withoutcommit, it also picks what to scan. If you leave it out, Gecko uses the repository’s default branch.- The commit must already be pushed. If it isn’t, the request fails with
400 commit_not_found. - A pinned scan is a deep scan of the repository at that commit, not a scan of the diff.
id. Poll GET /scans/{id} or subscribe to
webhooks to know when results are ready.
Scan several repositories
Sendrepository_ids instead of repository to cover up to 100 connected
repositories in one scan. Each request takes one of the two fields, not both.
Get the IDs from GET /repositories.
- Gecko creates one scan, and it counts once against your plan’s scan limit.
branchapplies to every repository. If you leave it out, Gecko scans each repository’s default branch.- You can’t send
commit,base_commit, orproviderwithrepository_ids. - Every ID must be a connected repository your API key can access. If one
isn’t, the request fails with
400 repository_not_connectedand Gecko creates no scan. - All repositories in a scan must use the same provider. A request that mixes
GitHub and GitLab fails with
400 mixed_repository_providers, so start one scan per provider. - The first ID is the scan’s primary repository, and the
repositoryfield in the response names it.
If a repository can’t be checked out, for example because it has no branch
matching
branch, Gecko skips it, scans the rest, and records the skip on
the scan.