Payout API
control checklist
All 46 recommended-action entries from CERT-In's two payout-API advisories, organised into eight control areas. Overlapping recommendations are retained so each advisory can be checked independently.
How to use this checklist
One row per CERT-In action, with the evidence beside it
- Each row carries the advisory it comes from: an id starting 46 is CIAD-2026-0046 of 9 October 2026, and an id starting 58 is CIAD-2026-0058 of 4 November 2026. The number after it is the action's position in that advisory.
- Mark each row V (verified), P (partially verified), G (gap) or N/A (not applicable). A control is verified only when there is evidence that it operates effectively; a policy or configuration alone may not establish that.
- Where both advisories recommend the same control, both rows are kept, because the November advisory usually adds a condition the October one did not have.
- Start with area 03. It holds the controls that decide whether a payout can be made without the loan behind it.
Payout API control checklist
CERT-In CIAD-2026-0046 (9 October 2026) and CIAD-2026-0058 (4 November 2026) · 46 recommended actions in 8 control areas · Security Brigade
Testing note: The examples describe checks for an authorised assessment. Active tests should follow a written scope and agreed rules of engagement, normally in a controlled test environment. Do not initiate live fund transfers or disruptive tests without the required approvals from the affected organisations and banking partners.
01 · 3 actions
API inventory and exposure
The entry points CERT-In lists are things nobody was watching: a deprecated API, an internal application reachable from the internet.
46-02
Is there a complete API inventory, and are deprecated and unused APIs formally decommissioned?
Evidence to review or test
The inventory with an owner per API; a comparison against observed traffic and gateway routes; decommissioning records.
46-16
Has internet exposure of internal applications been removed, with geofencing under formal change control?
Evidence to review or test
An external scan of your address space and DNS; firewall and gateway rules; the change records for each exposure.
58-10
Is there a rapid patching SLA for internet-facing applications, prioritising file-upload and deserialisation weaknesses, with WAF virtual patching as an interim control?
Evidence to review or test
The SLA; patch dates against disclosure dates for the last quarter; the virtual-patch rules in force.
02 · 4 actions
Caller identity
The fraudulent calls carried valid credentials and came from whitelisted infrastructure. Identity has to mean more than a source address.
46-01
Is MFA enforced for payment platforms, administrative portals, payment and bank APIs, and remote access?
Evidence to review or test
Identity-provider policy; a sample of accounts and service paths tested without the second factor.
46-03
Is per-session authentication enforced on all APIs, with long-lived static tokens eliminated?
Evidence to review or test
Token lifetimes in configuration; a replay test with an expired or reused token.
58-09
Are client applications treated as untrusted: every transaction and business rule validated on the server, requests signed with a nonce and timestamp, and app attestation and certificate pinning in place?
Evidence to review or test
A modified client build exercised against the API; a replayed signed request; attestation and pinning results.
58-13
Has IP allow-listing of shared third-party fetcher pools been replaced with mutual TLS, signed requests or scoped API keys on dedicated vendor endpoints?
Evidence to review or test
The allow-list reviewed entry by entry; which entries are shared pools; the identity control that replaced each.
03 · 8 actions
The payment journey
This is the centre of the attack: a debit call that succeeded with no loan, no sanction and no beneficiary check behind it.
46-04
Are multi-stage transaction flows bound cryptographically, so the debit-stage API cannot be invoked without the validation stages before it?
Evidence to review or test
A direct call to the payout API with the earlier stages skipped; the binding design and where it is verified.
46-05
Is there maker-checker or second-level verification for all fund movement?
Evidence to review or test
Approval records for a sample of transfers; an attempt to release a transfer with one identity.
46-06
Is every payout validated against a sanctioned or approved record, and is a reference identifier with no corresponding record rejected?
Evidence to review or test
A payout request carrying a fabricated loan reference; the validation logic at the payment stage.
46-07
Are per-transaction and per-day cumulative ceilings enforced at the API or payment stage?
Evidence to review or test
Requests above each ceiling sent directly to the API; where the limit is configured and who can change it.
46-08
Is beneficiary validation (penny-drop, name-to-account matching) applied, with velocity limits, on all disbursement paths?
Evidence to review or test
A transfer to an unvalidated beneficiary on each path, including refunds and bulk uploads.
46-15
Are behavioural anti-automation controls (rate limiting, sequence-binding) in place to disrupt script-driven API abuse?
Evidence to review or test
A scripted burst and an out-of-order call sequence against the payout endpoints.
58-17
Are transaction velocity rules, new-beneficiary cooling periods and amount-band anomaly detection applied across all payment and disbursement paths?
Evidence to review or test
Rule configuration per path; a new beneficiary used inside the cooling period.
58-18
Are server-side validation, maker-checker approval and reconciliation enforced for merchant settlement instructions submitted through payment aggregator APIs?
Evidence to review or test
A settlement instruction changed in transit; approval and reconciliation records for a sample.
04 · 4 actions
Secrets, keys, source code and build pipeline
A hard-coded token and a forged commit are both on CERT-In’s list of ways in.
46-10
Are all API keys and credentials stored in a secret management solution or HSM, with rotation enforced?
Evidence to review or test
Where each bank and aggregator credential lives; last rotation dates; who and what can read them.
58-11
Are hard-coded tokens and credentials prohibited in source and build code, with automated secret scanning across repositories and CI/CD, and short-lived, least-privileged tokens for build and deployment?
Evidence to review or test
Secret-scan output across repositories and pipeline variables; token scopes and lifetimes.
58-12
Is token use from unusual source IPs, user agents or geographies alerted on, and are branch protection, commit signing and mandatory code review enforced on production deployment branches?
Evidence to review or test
Repository settings for production branches; an unsigned commit attempt; the alert for an out-of-pattern token use.
58-19
For cryptocurrency custody: are wallet keys held in HSMs or cold custody, with multi-signature approval for withdrawals, anomaly monitoring and immediate rotation on suspected compromise?
Evidence to review or test
Applicable to digital-asset custody environments. Custody design; approval records for withdrawals; the rotation procedure.
05 · 11 actions
Environments, cloud and network
The November advisory describes entry through exposed UAT and staging systems, cloud management services, and unrestricted outbound traffic.
46-17
Are development, testing and production environments isolated?
Evidence to review or test
Network paths and credentials shared between environments; a reachability test from non-production to production.
58-06
Are UAT, staging and testing environments isolated from production, with no shared credentials, tokens, API keys, databases or direct network paths, and are internet-facing non-production environments hardened and monitored to production grade?
Evidence to review or test
The same test from each non-production environment; hardening and monitoring coverage of internet-facing ones.
58-07
Is any access from non-production systems to production systems, credentials or data alerted on and investigated?
Evidence to review or test
The detection rule and a triggered example with its investigation record.
46-09
Is outbound traffic from production servers restricted through allow-listing?
Evidence to review or test
Egress rules per server segment; an outbound connection to an unlisted destination.
58-14
Is default-deny egress enforced from server segments through allow-listed proxies?
Evidence to review or test
Proxy policy and logs; a direct HTTPS connection attempt that bypasses the proxy.
58-15
Is TLS inspection deployed, with alerting on self-signed certificates presented over non-standard ports?
Evidence to review or test
Inspection coverage; the alert rule and a test connection that should raise it.
58-03
Are cloud audit logs monitored centrally and in real time, with alerts on remote command execution, serverless code updates and the creation of roles, users or access keys?
Evidence to review or test
Audit-log pipeline; the alert for each event type, exercised in a test account.
58-04
Are cloud management and serverless execution restricted to approved principals and accounts, with alerts on interactive remote-management sessions and serverless code changes in production?
Evidence to review or test
IAM policies for management and serverless services; who holds them; the production alerts.
58-05
Are instance metadata protections and least-privilege roles enforced, wildcard egress denied in security groups, and serverless functions configured with no internet egress by default?
Evidence to review or test
Configuration review of instances, roles, security groups and functions in the payment estate.
58-08
Are application deployment directories read-only for the runtime user, with file-integrity monitoring for web shell detection?
Evidence to review or test
File permissions on application servers; a test file written to a deployment directory and the alert it raises.
58-26
Are zero-trust principles adopted: micro-segmentation, identity-based access for every user and service, and inspection of east-west traffic?
Evidence to review or test
Segmentation design against the payment estate; lateral movement attempted from one server segment to another.
06 · 9 actions
Monitoring, reconciliation and logging
Money moved in minutes, on holidays and weekends. Detection has to work when nobody is at a desk.
46-11
Are fund transfers monitored in real time with velocity, amount-band and new-beneficiary checks, and with automated blocking or holding?
Evidence to review or test
The rules and their thresholds; a test sequence that should be held, and whether it was.
58-16
Does monitoring flag micro-value test transactions to new beneficiaries followed by high-value transactions, and hold or review such flows?
Evidence to review or test
A one-rupee transfer to a new beneficiary followed by a larger one in a test environment; the alert and the hold.
46-12
Is there near-real-time reconciliation of system records with bank and aggregator statements?
Evidence to review or test
Reconciliation frequency; how a transfer with no matching record surfaces, and how long it takes to reach a person.
46-13
Is logging centralised, with retention of at least 180 days?
Evidence to review or test
Log sources for the payment estate; retention settings; the oldest retrievable event.
46-14
Is EDR with tamper protection deployed on servers and endpoints?
Evidence to review or test
Coverage of payment servers; tamper-protection settings; an attempt to stop the agent.
58-01
Is there heightened 24x7 monitoring, with increased staffing and lowered alert thresholds during holidays, weekends and other low-coverage windows?
Evidence to review or test
The roster and thresholds for the next holiday; response times to out-of-hours alerts over the last quarter.
58-23
Is SIEM correlation tuned for machine-speed, multi-technique attack patterns, with automated (SOAR-based) containment?
Evidence to review or test
Correlation rules for parallel and adaptive attack activity; the containment playbooks and their triggers.
58-24
Are deception measures deployed: honeypots, decoy administrative panels and canary credentials?
Evidence to review or test
The decoys in place and the alert each raises when touched.
58-25
Are web roots, temporary directories and staging paths monitored for bursts of newly created scripts and rapid variants of the same script?
Evidence to review or test
The monitoring rule; a burst of test files and the alert it raises.
07 · 4 actions
Tool and AI governance
CERT-In names publicly available AI-driven offensive tools used in the campaign, and asks that the same tools are not running unnoticed inside.
58-20
Is the installation or execution of unauthorised security and attack tools prohibited on corporate infrastructure, and are their download and AI coding-agent processes with permission checks disabled monitored for?
Evidence to review or test
The policy; endpoint and proxy rules that enforce it; a detection test.
58-21
Is LLM API usage monitored and governed, including LLM API traffic from servers or other non-development hosts?
Evidence to review or test
An inventory of LLM API keys; egress logs for LLM endpoints by source host.
58-22
Does a policy require written approval, a defined scope and a time window for all penetration testing?
Evidence to review or test
The policy and the authorisation record for each test in the last year.
46-18
Is VA/PT conducted regularly, including API security testing?
Evidence to review or test
Reports for the last cycle, the APIs each covered, and the closure status of their findings.
08 · 3 actions
Incident response
Organisations that had already remediated were attacked again. Containment and eradication have to be verifiable.
46-19
On detection of suspicious activity: are compromised accounts disabled, affected APIs suspended, forensic evidence preserved, all credentials rotated, and CERT-In/CSIRT-Fin notified?
Evidence to review or test
The playbook; who can suspend a payout API out of hours and how; the last exercise or incident record.
58-27
On detection of suspicious activity: are affected hosts isolated, forensic evidence and logs preserved, all credentials rotated, and CERT-In/CSIRT-Fin notified?
Evidence to review or test
Host isolation procedure and tooling; evidence-handling steps; reporting contacts held by the on-call team.
58-02
After any incident, is persistence assumed, with full eradication, forensic verification and continued periodic threat hunting?
Evidence to review or test
The eradication report for the last incident; the hunting schedule and its findings.
Start with the controls that authorise the money.
A payout should not succeed merely because the calling server and credentials are trusted. The Payout API Security Audit examines the sanction, beneficiary, approval and transaction-limit controls that should prevent an unauthorised transfer.