Settings

Private account settings, research controls, team access, and transparent capability limits.

Safety enforcement

Scope Guard and Compliance cannot be disabled. Live network scanning and automatic disclosure are not available.

Workspace jobs per minute · maximum 60Response captures per jobResearch enabled

Your researcher profile

Profile details are private to your account. Roles are assigned by the owner, never by this form.

Staff access

Owner-managed login accounts. Researchers can save drafts; only an administrator can verify scope, run jobs, change controls, or approve reports.

Storage & capabilities

Research records use the app’s built-in key-value store with account-private scope. Authentication, jobs, realtime, AI, and public policy uploads use the platform SDKs.

A relational database is a Pro feature. Current records are not SQL tables. Upgrade to Pro for a managed database, or connect your own Supabase project for free from the Database panel.

AI advice uses supplied context, not live web search. Arbitrary external scanning, TLS probes, test-account IDOR validation, advisory lookups, and automated disclosure are not implemented.

Uploaded policy files are public. Evidence and reports remain in account-private records; do not put confidential research in public file storage.

Local core verification

34 / 35 passed

These deterministic checks execute locally on first paint and can be rerun. They test shared safety logic, evidence rules, role gates, and reports—not cloud authentication, background execution, live scanning, or production readiness.

Show test results

Exact allowlisted asset is permitted

Unknown scope is blocked

Out-of-scope asset is blocked

Related subdomain is never inherited

Explicit deny overrides allowlist

Unverified policy cannot authorize

Ambiguous policy blocks testing

Expired authorization blocks testing

Missing permitted method blocks testing

Emergency stop blocks research

HTTP target is rejected

Private IP literal is rejected

Credentials in target URL are rejected

Query strings are not silently widened

Wildcard assets require clarification · Expected operation to be blocked

Path trailing slashes are not silently widened

Evidence must match exact authorized asset

Request credentials cannot be imported

Cookie values are redacted before saving

Header newline injection is rejected

Findings remain unverified by default

Excluded categories are not emitted

Strong header capture suppresses unsupported findings

Wildcard CORS does not claim authenticated exploitation

Report includes all thirteen required sections

Unverified finding cannot be approved

Verified finding still needs a review rationale

Verified review can be approved for manual export

HTML exports escape active markup

Anonymous caller cannot mutate records

Researcher cannot approve or scan

Owner role is decided by server context

Email text never grants administrator rights

Request rate limit blocks excess jobs

Inspection is deterministic and deduplicable