Most endpoints are org-scoped under /api/v1/orgs/:orgId/... and return { data, meta } responses. Authenticate with your session cookie (browser) or an API key sent as a Bearer token — create keys under Settings → Security → API keys in the dashboard.

API key scopes

Every API key carries a set of API key scopes, chosen when the key is created and fixed from then on. The same scopes apply to the REST API and to the MCP endpoint: a key with findings:read can read findings over either, and nothing it was not granted over either. A new key must have at least one scope; the create request is refused with 400 otherwise.

On REST, a read (GET) needs the area's :read scope; any other method needs its :write scope. A write scope does not grant reads. A request the key's scopes don't cover is refused with 403 and a body naming the missing scope: { "error": "API key lacks the required scope 'findings:write'", "requiredScope": "findings:write" }.

Routes not listed below are refused for scoped keys, with requiredScope: null. That is deliberate for organisation administration (API keys, MCP clients, team, settings, support access), in-app notifications (each person's own inbox), security questionnaires and the trust centre. A dash means no scope grants that kind of access: for example, control status, compliance targets and assessments, and creating or editing systems are changes a person makes, not a key. Each operation in the reference below also states its scope.

Path (under /api/v1/orgs/:orgId) Read Write
/findings findings:read findings:write
/findings/synthesize-actions — actions:write
/findings/*/severity-override — —
/actions actions:read actions:write
/incidents incidents:read incidents:write
/risk-register risks:read risks:write
/policies policies:read policies:write
/frameworks/*/link-policies — policies:write
/compliance controls:read —
/compliance/controls/*/evidence controls:read evidence:write
/runs runs:read runs:write
/agent-jobs runs:read runs:write
/agent-jobs/active — —
/scopes runs:read —
/scopes/*/runs runs:read runs:write
/scopes/*/config — —
/scopes/*/exclusions connectors:read connectors:write
/scopes/*/sbom assets:read —
/reports reports:read reports:write
/assets assets:read assets:write
/connectors connectors:read connectors:write
/dashboard findings:read —
/dashboard/summary findings:read reports:write
/dashboard/ai-usage — —
/my-work — —
/digest findings:read —
/identities assets:read —
/audit-logs audit:read —
/pentests findings:read findings:write
/vendors risks:read risks:write
/campaigns campaigns:read campaigns:write
/campaigns/* campaigns:read —
/api/webhooks/aws — connectors:write

The most specific path wins, so control evidence uploads need evidence:write although the rest of /compliance is read-only to keys. Control status stays out of reach for keys: controls:write exists for MCP only, is granted to signed-in users' clients, and cannot be put on an API key.

Some responses are narrower for an API key than for a signed-in person, matching what the scope grants on MCP: audit log events leave out IP address and user agent, and agent jobs leave out the job result and platform ids. To see active agent jobs, list /agent-jobs?status=active; the /agent-jobs/active listing also reconciles jobs, so it is not available to keys.

Keys without scopes — keys created before scopes were required — are refused on every REST route with 403, and have no MCP access. They are marked in the dashboard's key list. Create a replacement key with the scopes your integration needs and revoke the old one.

Support keys — keys issued for Quant Support — work on MCP only. Every REST request made with one is refused with 403, whatever scopes it holds. See Support keys.

Connecting an AI client such as Claude or Cursor? Use the Model Context Protocol integration instead of calling the REST API directly.