Sonar documentation
Sonar organizes authorized infrastructure into workspaces. A workspace holds one or more domains or public IPs (its assets) with their scans, findings, members and webhook. Your account holds every workspace you belong to; workspaces can be renamed, merged, or deleted at any time, and a domain counts as a single asset no matter how many subdomains are observed beneath it. How many assets, workspaces and people you can have depends on your plan (see Plan & usage in the console); assets count from the moment they are added, verified or not.
Add and verify an asset
- Sign in with Discord or Google. New accounts start in account mode; open Workspaces.
- Name the workspace and paste one or more domains or public IPs — one per line. Each becomes an asset in the new workspace.
- For domains, publish the displayed TXT value at the displayed challenge name. For IPs, serve the challenge at the displayed HTTP URL without a redirect.
- Select Verify ownership for each asset. Once verified, the challenge is hidden — the card keeps a Remove action only. Scans recheck ownership every run, so leave records in place.
- Add more assets at any time from the workspace's asset panel, and set a scan schedule on the Scans page.
How a run works
One run covers every verified asset in the workspace. A run you start by hand counts as one manual scan against the workspace owner's daily allowance, shared across all of their workspaces and reset at midnight UTC; scheduled runs don't count against it. Each asset passes through:
- Discovery — subdomains from certificate transparency, the live certificate's SANs, public web archives and upstream subdomain feeds; names in the domain's DNS records (MX, NS, SPF, DMARC, SRV and CNAME); names referenced by its pages, security headers, robots.txt, sitemaps and scripts; a curated list of about 900 common names; and permutations and one level of recursion beneath the names that exist. Wildcard DNS is detected so guessed names are not mistaken for hosts. Other domains that carry your brand and appear in DNS or on your pages are offered as suggestions, never scanned. Each workspace tracks up to 500 hosts.
- Ports — a paced TCP connect sweep across your chosen coverage: common (228 ports) or extended (1,504 curated ports) at no more than 6 probes per second, or wide (ports 1–20,000 plus curated ports above 20,000, on Pro and above) or full (all 65,535, on Business) at no more than 25 per second, a low-intensity version fingerprint pass on open ports, and, on privileged deployments, single UDP probes to DNS, NTP and SNMP. A sweep that times out is reported as incomplete rather than closed, and an address that answers on every port is flagged instead of producing port findings.
- Web — HTTP(S) analysis, TLS certificate and cipher analysis, security-header and version-disclosure review, technology fingerprinting, favicon fingerprinting, about 60 read-only exposure, debug-endpoint and admin-panel signatures, and one latest screenshot of the web surface. The signatures also run on up to four other open web ports per host (such as 8080 or 8443).
- DNS & mail — SPF, DMARC, CAA and DNSSEC hygiene, plus deeper mail-security checks for MTA-STS, TLS-RPT, BIMI and DKIM selectors.
- Domain — registration expiry and transfer lock from public RDAP records, and registered lookalike (typosquat) domains, checked by DNS only, for verified registered domains.
- Exposure — read-only checks for accidentally published files, a signature template engine (additional exposure and login/admin-panel and API-surface detection), and secret-pattern scanning of the page and its own same-origin scripts.
- Change & takeover — subdomain-takeover confirmation against the hosting provider's own response, and per-change alerts summarizing opened and closed ports since the last observation.
- Known vulnerabilities (CVEs) — correlates the services and software fingerprinted on a host (fingerprinted server software by version range, service and OS components, and versioned client-side libraries, each checked against upstream vulnerability databases) to published CVEs, raising tracked findings prioritized by known-exploited status and exploit-probability scores, with severity-based remediation SLAs. A CycloneDX SBOM can be uploaded per asset for dependency-level CVEs.
- Attribution discovery — surfaces related assets you never named (hosts sharing a verified origin IP, and sibling domains on your certificates) as suggestions to review, add and verify.
- CDN & origin mapping — detects Cloudflare, Akamai, Fastly, CloudFront and other CDN/WAF edges (by IP range and CNAME), tests those hosts at the web layer instead of port-scanning shared edge infrastructure, and flags hosts that resolve directly to origin and bypass the CDN protecting their parent domain.
A safe activity log records each milestone; open View log on any scan.
Organize your inventory
Tag assets with lowercase labels to group them, and filter by tag:. Every asset carries a lifecycle state — new, active, changed or decommissioned; the worker marks discovered hosts "new" and flips a settled asset to "changed" when its fingerprint moves, and decommissioning also removes it from scanning. Star assets to build a personal watchlist, save any asset search as a named view, and use the command palette (⌘K / Ctrl-K) to jump between views or run a search from anywhere.
Risk, priority and SLAs
Every asset carries a 0–100 posture score and letter grade derived from its open findings, and the workspace Overview shows the daily score trend. Each finding also gets a priority (0–100) computed from its severity, whether the CVE is known to be exploited in the wild, its exploit-probability score (EPSS), and the asset's business criticality — the findings queue sorts by priority by default. Severity-based remediation SLAs mark overdue findings. Known-exploited and exploit-probability data come from public upstream feeds and refresh in the background.
Findings workflow
Assign a finding to a teammate (they are notified directly), change the status of many findings at once, and @mention members in comments. Suppression rules keep known-accepted findings out of the open queue — a matching finding is recorded as a false positive now and on every future scan instead of reopening. Re-test an asset on demand to queue a fresh scan of its scope. An activity feed records assignments, status changes, comments, re-tests and detected changes. Each member chooses which notification categories they receive.
Investigate
Asset search matches hostnames and collected evidence. Select an asset to open its drawer: a header with tags, lifecycle, a watch toggle and a re-test button, then tabs for Overview (evidence, technology stack, services, UDP, screenshot, activity and history), Findings, and Advanced (software inventory and authenticated read-only checks). Findings include severity, priority, known-exploited and exploit-probability data, evidence and remediation. Resolved findings reopen if observed again, unless suppressed.
Search syntax
Combine terms with spaces to require every condition. Supported fields: host, ip, port, udp, http.status, cert.issuer, service, tech, severity, excluded, tag, lifecycle, watched, nettype and favicon. The nettype: filter matches a host’s infrastructure classification — tor-exit, vpn, proxy, datacenter, mobile, residential, bot or cloud. Quote phrases containing spaces. Ports, IPs and favicon hashes match exactly; hostnames, service fingerprints, technologies and issuers use text matching. Severity matches open findings.
port:443 http.status:200 cert.issuer:"Example Authority" tech:nginx severity:high tag:production lifecycle:new watched:true udp:161 host:api excluded:false
Attribution discovery
From Company → Discovery, find candidate domains that mention a brand label in public certificate transparency, and mobile apps from the public App Store. Results are candidates for review only — nothing is added to scope or scanned, and ownership must still be verified before an asset is added.
Public passive search — the Tohou Index
Public record search is its own product at search.tohou.io, signed in with the same account as this console. An API key with the index:read scope calls it programmatically at https://api.tohou.io/v1/index?q=example.com. Searches on the website and through the API share one daily Index allowance (10 a day on Free, 100 on Starter, 500 on Pro and 1,000 on Business); an API lookup also counts as one API request. Domain searches return public certificate-transparency names and certificate inventories, current DNS records, public web observations and WHOIS/RDAP registration. Address searches return open-port records with service details, hostname and reverse-DNS records, network and ASN data, registry and geolocation information, and WHOIS registration. ASN searches return announced prefixes. Domain results add historical URLs from a public web archive; both domain and address results include a live observation — a bounded direct check of common service ports, HTTP(S) status and the TLS certificate — but only for targets inside your own verified scope. Every other lookup is answered from third-party public records without sending traffic to the target. Index lookups are stateless: no search history is stored and nothing is added to workspace inventory.
Authenticated read-only checks
Configure exact HTTPS paths with a dedicated low-privilege bearer token or session cookie. HEAD checks collect headers. GET checks require you to confirm the path is read-only, compare anonymous and authenticated status codes, and discard the response body after at most 16 KiB. An anonymous 2xx result is a review signal, not proof of an access-control vulnerability. Sonar follows no redirects.
Authorized intrusive testing (opt-in)
Separately from the observation above, workspaces authorized to actively test their own assets can enable a bounded, non-destructive active-check tier. It is off by default and gated three ways: the operator must enable it for the deployment, a workspace administrator must turn it on, and each asset requires its own explicit written consent. Only an asset meeting all three conditions is tested. After a shallow same-origin crawl (forms, any published OpenAPI spec and a headless browser pass for single-page apps), the checks are detection-only: reflected XSS, error-based, boolean and time-based blind SQL injection, open redirect, path traversal, server-side template injection, OS command injection (a benign echo marker), CRLF injection, cloud-metadata SSRF, CORS and host-header misconfiguration, cookie flags, exposed files and consoles, GraphQL introspection and JWT review. Findings are per endpoint and parameter with a confidence rating. An optional encrypted login recipe (bearer token, cookie, header or a test-account form login) lets the same checks run behind authentication. They extract nothing beyond a single proof marker, change no state, send only GET and read-like POST requests and never guess credentials. See the scanning policy for the full boundaries.
Integrations, tickets and cloud connectors
Workspace administrators can connect Slack, Discord, generic webhooks, Microsoft Sentinel and Azure Log Analytics for notifications, and Jira Cloud, Linear, GitHub Issues and ServiceNow for tickets. A ticket is opened once per finding and closed when the finding is resolved. Credentials are encrypted and never shown again. Cloud connectors (AWS, Google Cloud, Azure and Cloudflare) use read-only credentials to list public IPs, hostnames and buckets, raise misconfiguration findings and suggest assets. Only resources the provider itself proves you control can be approved without DNS verification, and that proof lapses if the connector stops listing the resource or fails to sync for 48 hours. Cloud connectors are included in Business and Enterprise workspaces. Pro, Business and Enterprise workspaces can also run customer-hosted agents that scan declared private ranges from inside the network; the agent only connects out and refuses non-private addresses.
Emerging threats
Sonar keeps a catalogue of newly exploited vulnerabilities and matches it against recorded software and versions without sending traffic. Where the version is not visible the result is marked tentative. New entries in the known-exploited list are matched the same way and trigger a targeted re-check of only the affected assets where the plan has scheduled scans. A vendor-path probe runs only under the intrusive testing gates above.
IP intelligence
Every address a scan resolves to, and every address shown in the Index, is classified from public reputation feeds: Tor exit node, VPN, open or anonymising proxy, datacenter or hosting, cloud provider, abuse/bot infrastructure, and — from the OpenASN dataset — residential or mobile. Classification is passive; it reads published lists, not the host. A verified asset that resolves to a Tor exit, VPN, proxy or abuse feed raises an informational finding, since it often means shadow IT or a compromised host relaying traffic. Filter inventory with nettype: and look up any address at GET /v1/ip-intel?ip=…. Residential and mobile detection can be sharpened with an optional provider key; nothing paid is called without one.
Scheduling and notifications
Each workspace chooses Manual, Daily, Weekly or Monthly runs with a count per period; gaps are spread evenly across UTC periods and shown before saving. How often scheduled runs can happen depends on the owner's plan: weekly on Free, daily on Starter, twice a day on Pro and four times a day on Business. Scheduled runs don't use the manual-scan allowance, and a workspace runs one scan at a time. Add a Discord or Slack webhook when creating or editing a workspace, or any destination under Integrations (see above), to receive formatted finding and scan updates; delivery failures retry with increasing delays and are reported in console notifications.
Roles
Owners manage all workspace settings, members and deletion. Administrators manage assets, keys and analyst invitations. Analysts investigate findings, export reports and run scans. Only the platform operator can change capacity, suspend accounts or ban users.
Reef: threat intelligence from our own sensors
Tohou operates inbound sensors on its own addresses — hosts that accept unsolicited connections and record what the sender tried. They never initiate traffic, never scan and never attack back. What reaches them shows which vulnerabilities are under active exploitation, often before a CVE reaches any known-exploited list.
A vulnerability is only called “under active attack” when at least three distinct addresses attempted it in 24 hours and the request matched a signature specific enough to confirm the CVE, not merely a scan for a common path. Where Sonar has recorded an affected version on one of your assets, the finding is raised and its severity lifted, with the observation attached as evidence.
Published entries carry a confidence tier, age out on their own and stop being published once the activity stops. Addresses on mobile and shared networks are capped at a lower tier, because one address there can carry thousands of unrelated people. An address is never described as malicious, and an address with no record is not thereby safe. To request removal, write to [email protected] from the address’s abuse contact; we ask for proof of control, decide within thirty days, and a grant stops publication within one cycle without stopping the sensors, so renewed attacks are published again. The privacy policy has the full route and the retention periods.
The public summary is at tohou.io/reef. Every API endpoint needs a key: GET /v1/reef/ip/{address} for one address, GET /v1/reef/exploited for what is being exploited now, GET /v1/reef/feed for a paged bulk export and GET /v1/reef/stream for live observations as they arrive. Reef is not included on the Free plan and starts at Starter; the bulk feed needs Pro and the live stream Business. The feed and the stream are key-only, one live stream at a time per account. A Reef read inside an Index lookup costs nothing extra. The pricing page lists the daily allowance for each plan.
Single sign-on and SCIM
Workspaces on the Business plan and Enterprise plans can sign in through their own OpenID Connect provider (Okta, Microsoft Entra ID, Google Workspace, Auth0, Keycloak). In Settings, add your email domains and publish the DNS TXT record shown for each, then enter the issuer URL, client ID and secret, and register the redirect URI Sonar displays. Sign-in uses authorization code with PKCE and verifies the signed ID token against your provider's published keys. New people are provisioned just in time with the default role you choose. A verified domain routes sign-in by email address and lets Sonar link an existing account with the same address; people on a domain that is not verified cannot sign in. Only owners can make administrator the default role, and only owners can change the issuer, client ID, enforcement, domains or SCIM token. SSO sessions last 12 hours by default (1 to 720, set in Settings) before the person signs in at your provider again; SCIM deactivation ends them at once. Only verified email addresses are ever used to link accounts. SAML is not supported.
"Require SSO" is available once a domain is verified. Members other than the owner then reach the workspace only through SSO; the owner keeps a normal sign-in as emergency access, and existing API keys keep working until the person is removed. Directories can create and deactivate people through SCIM 2.0 at /scim/v2/Users with the workspace's SCIM token, for addresses on verified domains. Deactivating removes workspace access, keys and sessions immediately, and never touches the person's other workspaces.
Compliance evidence
The Compliance page maps findings to controls in SOC 2, ISO/IEC 27001:2022, PCI DSS 4.0, NIST CSF 2.0, the HIPAA Security Rule, CIS Controls v8 and NIS2, and exports the result as a print-ready report, CSV or JSON (GET /v1/compliance/{framework}). Each control is shown as no exposure found, at risk or not assessed. Sonar sees only what is reachable from the internet, so this is supporting evidence for your auditor, not a certification, and controls about internal systems, people or process are listed as not assessed. Sonar is not a PCI Approved Scanning Vendor.
API authentication
Create keys in the console under Account → API keys. There is one key type, ts- followed by 32 characters, owned by your account and shown once at creation. Each key has scopes: index:read for Public Index lookups, and workspace scopes (assets:read, assets:write, findings:read, findings:write, scans:read, scans:write, reports:read) that act on one workspace you choose, with your role there. Send the key in the Authorization header. The API hostname is api.tohou.io; endpoints use the /v1 prefix. Requests return JSON unless an export format is selected.
curl https://api.tohou.io/v1/assets -H "Authorization: Bearer $TOHOU_API_KEY" curl "https://api.tohou.io/v1/index?q=8.8.8.8" -H "Authorization: Bearer $TOHOU_API_KEY"
API access depends on your plan: Free has none, Starter has read-only access (GET requests and read scopes only), and Pro, Business and Enterprise have full access. Workspace scopes also need API access on the plan of the workspace's owner. Every key request counts against your account's daily API quota (150 on Starter, 500 on Pro, 1,500 on Business), which resets at midnight UTC; an Index lookup also uses one of the Index searches you share with the website, and is refused without using either if one of the two is exhausted. Requests are also limited to 120 a minute per account. Never include API keys in browser URLs or public repositories. Invalid keys return HTTP 401. Missing permissions, or a plan without the needed access, return HTTP 403. Limits return HTTP 429.
The Open CVEs endpoints need an API key, but any key works and so does any plan, including Free: 50 requests a day on Free, 1,000 on Starter, 5,000 on Pro and 20,000 on Business, counted separately from your API and Index allowances, at up to 120 requests a minute per account. The pages at tohou.io/open-cves stay public and need nothing at all.
curl https://api.tohou.io/v1/cves/CVE-2024-3400 -H "Authorization: Bearer $TOHOU_API_KEY"
| Endpoint | Permission | Purpose |
|---|---|---|
| GET /v1/assets?q=… | assets:read | Search inventory, 50 rows per page; use offset |
| GET /v1/assets/{id} | assets:read | Asset detail, tags, findings and recent states |
| GET /v1/scopes | assets:read | List assets in the workspace |
| POST /v1/scopes | assets:write + admin | Add a domain or public IP |
| POST /v1/scopes/{id}/verify | assets:write + admin | Check ownership |
| POST /v1/assets/{id}/rescan | scans:write | Queue a re-test of this asset's scope |
| GET /v1/scans | scans:read | List scan runs with progress and activity logs |
| POST /v1/scans | scans:write | Queue a run for one asset or the whole workspace |
| POST /v1/scans/{id}/cancel | scans:write | Cancel a run |
| GET /v1/findings?sort=priority | findings:read | List findings with priority, known-exploited status and exploit probability |
| PATCH /v1/findings/{id} | findings:write | Set status, assignee, dueAt, comment |
| PATCH /v1/findings/bulk | findings:write | Set status on many findings at once |
| GET/POST /v1/suppressions | findings:read / write | List or add suppression rules |
| GET /v1/reports?format=json | reports:read | Export assets and findings |
| GET /v1/reports?format=csv | reports:read | Export findings as CSV |
| GET /v1/index?q=… | index:read | Index lookups for domains, addresses and ASNs |
| GET /v1/reef/ip/{address} | reef:read | What Tohou's sensors have seen from one address |
| GET /v1/reef/exploited | reef:read | Vulnerabilities being exploited against the sensors right now |
| GET /v1/reef/feed | reef:read (Pro and above) | Bulk export of observed addresses, paged |
| GET /v1/reef/stream | reef:read (Business and above) | Live stream of new observations |
| GET /v1/ip-intel?ip=… | index:read | Classify an address: Tor exit, VPN, proxy, datacenter, cloud, residential, mobile or abuse/bot |
| GET /v1/emerging | findings:read | Emerging-threat catalogue with this workspace's affected assets |
| POST /v1/emerging/rescan | scans:write | Re-check only the affected assets |
| GET /v1/assets/{id}/web-testing | assets:read | Per-endpoint web-test results and login-recipe status (no secrets) |
| GET /v1/findings/{id}/tickets | findings:read | Tickets opened for a finding |
| GET /v1/suggestions | assets:read | Discovered candidate assets with source and confidence |
| GET /v1/compliance | reports:read | Compliance summary for every framework (plan feature) |
| GET /v1/cves | any key | Open CVEs: exploited now, just published, most likely next |
| GET /v1/cves/{CVE-ID} | any key | One CVE: severity, exploit probability, exploited status, affected software |
Queue a scan
curl https://api.tohou.io/v1/scans -H "Authorization: Bearer $TOHOU_API_KEY" -H "Content-Type: application/json" --data '{"scopeId":"your-verified-scope-id"}'Omit the scope ID to queue one run covering every verified asset in the workspace. A queued scan is not a completed scan; poll the scan list to inspect status, progress and the activity log.
Coverage and boundaries
Sonar performs real-time, bounded observation of verified scope only — by design, not a worldwide pre-built search index. Outside the opt-in, consent-gated intrusive tier described above, it performs no exploitation, credential attacks, denial-of-service testing, or decoy or spoofed-source scanning. Cloud connectors are read-only. Exact behavior is defined by the scanning policy.
Changelog
Recent additions to the platform. Dates are approximate and features roll out behind the normal workspace controls.
- Sign-in — Google sign-in alongside Discord.
- Authorized intrusive testing — an opt-in, consent-gated tier of bounded, non-destructive active checks.
- Discovery — attribution discovery (a company's domains and networks from public registry and certificate records, candidate domains by brand name, and mobile apps), exposed-secret scanning of a site's own scripts, API-surface detection, and
favicon:pivots. Company results are candidates: domains are verified before scanning, and mapped IP ranges stay passive until written authorization is reviewed. - Detection depth — technology fingerprinting, deeper mail security (MTA-STS, TLS-RPT, BIMI, DKIM), a read-only signature template engine, per-change alerts, and configurable port coverage.
- Vulnerability intelligence — known-exploited and exploit-probability enrichment from upstream feeds, risk-based finding priority, and remediation SLAs.
- Workflow — finding assignment notifications, bulk actions, suppression rules, re-test on demand, @mentions and an activity feed, and per-user notification preferences.
- Organization — asset tags, lifecycle states, watchlists, saved searches, and a
⌘Kcommand palette.
Support
Contact [email protected] for account and product support. Report unexpected scanning to [email protected] with the target and UTC timestamp.