API penetration testing from Sumrite for the APIs that power your product.
Most modern apps are a front end on top of APIs. If endpoints leak data or skip permission checks, every customer is at risk. We test APIs the way attackers abuse them — then deliver a clear report.
What this helps you achieve
Stop cross-customer data leaks
Find broken access rules before one user reads another company's records.
Ship integrations with confidence
Validate partner connections and mobile backends before production traffic.
Pass enterprise security reviews
Show auditors structured API testing — not a generic scan export.
Reduce fraud and abuse
Catch weak tokens, missing rate limits, and logic flaws that enable takeover.
Sumrite’s testers run your API security testing by hand, and the report is issued through a CERT-In empanelled auditor. One team from scoping to sign-off, built for Indian regulators, enterprise buyers and government work.
Accepted where it is requiredRegulators, auditors and procurement teams that ask for a CERT-In empanelled auditor’s report, including SEBI-regulated entities and government and PSU contracts.
One team, one contractSumrite handles scoping, testing, the report and the retest. You do not manage a second vendor.
Real testing, not a scanManual exploitation with evidence for every finding, and a retest after your fixes.
Yes. We run the full engagement for you and route the formal attestation through our network of CERT-In empanelled partner auditors, so you deal with one team from scoping to report.
What is API security testing?
API security testing is an expert-led assessment of the endpoints your web app, mobile app and partners call behind the scenes. Most serious API flaws are not visible in the interface: the screen only shows your own records, but the endpoint behind it may return anyone's if the ID in the request is changed. A tester maps the endpoints, signs in as different users and roles, then replays, alters and reorders requests to see what the server allows. The work follows the OWASP API Security Top 10, with particular attention to broken object-level authorisation, broken authentication, excessive data exposure and missing rate limits. REST and GraphQL are both covered. Each confirmed issue comes with the exact request that proves it, its severity and the server-side fix. For a scoped API, Sumrite delivers the report within 48 hours of testing starting, with re-testing of fixes included.
What are BOLA and BFLA?
Broken object-level authorisation (BOLA): the API returns or changes an object, such as an invoice or a user record, without checking that it belongs to the caller. Changing an ID in the request is enough to read another customer's data.
Broken function-level authorisation (BFLA): an ordinary user can call a function meant for administrators or another role, such as an admin endpoint or a bulk export, because the API checks that the caller is logged in but not what they are allowed to do.
How Sumrite API testing maps to the OWASP API Security Top 10 (2023)
OWASP API risk
What Sumrite tests
API1 Broken object level authorization
Every object identifier swapped across users and tenants
Whether changing an ID returns another user's or tenant's data
GET /orders/1043 returns a different customer's order
Authentication
Token issue, expiry, refresh, revocation and key handling
A JWT accepted after the user has logged out
Function-level access
Admin and internal endpoints reachable by ordinary users
A normal account can call the user-delete endpoint
Data exposure
Fields returned that the client never displays
Password hashes and emails in a list response
Abuse controls
Rate limits, enumeration and resource exhaustion
Unlimited one-time-code attempts on login
How an engagement runs
Scope and NDA. Tell us what needs testing and why. You get a fixed-price quote and an NDA is signed before any technical detail changes hands.
Rules of engagement. Targets, test windows, accounts, safety contacts and anything off-limits are agreed in writing before testing starts.
Expert-led testing. A researcher tests by hand, using tools to speed up the routine parts. Every finding is reproduced and confirmed before it is reported.
Report. An executive summary for leadership, then each finding with a CVSS severity, the evidence, steps to reproduce and a fix your team can apply.
Re-test. Once fixes are in, we test them again and record the result, so the final report shows what was found and what is now closed.
API Security Testing: common questions
What do you need to start an API test?
A base URL for a test environment, documentation if it exists (OpenAPI, Postman or GraphQL schema), and at least two accounts per role so cross-user access can be tested. Without documentation we map the API from the app's traffic.
Yes. GraphQL testing adds introspection, query depth and batching abuse, and field-level authorisation on top of the usual authentication and access checks.
Testing is done against agreed accounts and data, preferably in staging. Destructive actions are only performed on records we create, and the rules are agreed in writing first.
We run the full engagement for these frameworks, and where a formal audit or attestation is needed we route it through our local partner firms, so you deal with one team.
Every phase is run by a security expert, with tools used only to speed up the routine parts. Findings are reproduced and confirmed before they reach your report.
Discovery
Endpoint and parameter enumeration, including legacy versions
Schema introspection and shadow-API discovery
Authorisation
Object-level checks on every identifier (BOLA)
Function-level: calling admin endpoints as a normal user
Property-level exposure and mass assignment
Authentication and tokens
JWT and OAuth weaknesses: weak secrets, scope abuse
API key handling and rotation
Rate-limit and brute-force controls
Data exposure and injection
Excessive data returned in responses
Injection in JSON and GraphQL, including batching and depth abuse
SSRF through URL parameters and webhooks
Business-flow abuse
Abusing sign-up, checkout and voucher flows at scale
Replay and race conditions
◆ Expert-only work: beyond scanners and AI
Working out which objects belong to which tenant, then crossing the line
Chaining a small data leak into account takeover
Reasoning about authorisation field by field in GraphQL
Abusing intended features in ways the spec never imagined
Step 3 of 603
Draft report in 48 hours
Findings are written for two readers at once: developers who need the exact fix and auditors or buyers who need confidence.