Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-115332 Overview
CVE-2026-115332 is a SQL injection vulnerability affecting Lotus through version 0.10.2, an open-source pricing and billing platform. Authenticated organization users can inject SQL by registering crafted subscription_filter_keys that bypass the application's keyword-only validation. When an attacker creates a subscription using the malicious key, the value is rendered as an unquoted group-by identifier inside usage aggregation queries. Those queries run against the shared multi-tenant database, exposing data belonging to other tenants. The weakness is classified as CWE-89: Improper Neutralization of Special Elements used in an SQL Command.
Critical Impact
Authenticated tenants can read, modify, or exfiltrate records from the shared multi-tenant database, breaking tenant isolation and compromising confidentiality, integrity, and availability of billing data.
Affected Products
- Lotus (uselotus/lotus) versions up to and including 0.10.2
- The metering_billing backend component responsible for subscription filter serialization
- Deployments that expose the Lotus API to authenticated organization users on a shared multi-tenant database
Discovery Timeline
- 2026-10-22 - CVE-2026-115332 published to NVD
- 2026-10-22 - Last updated in NVD database
Technical Details for CVE-2026-115332
Vulnerability Analysis
The flaw lives in Lotus's subscription filter handling inside backend/metering_billing/serializers/model_serializers.py. The serializer validates that submitted filter keys match an allowed keyword pattern, but that check only screens the key syntax. It does not treat the stored value as untrusted input when the key is later interpolated into a raw SQL GROUP BY clause during usage aggregation.
Because Lotus serves multiple tenants from one database, an authenticated organization user can register a crafted subscription_filter_key, then create a subscription that uses it. The usage aggregation pipeline reads that key and concatenates it into the generated query as an unquoted identifier. Injected SQL executes with the application's database privileges, allowing cross-tenant reads and writes against billing, metering, and customer records.
Root Cause
The root cause is improper neutralization of a user-controlled identifier. The validation layer accepts strings that look like keywords, but the query construction layer trusts those strings as safe SQL fragments. There is no identifier quoting, allow-list enforcement, or parameterization at the point of use. See the vulnerable region in backend/metering_billing/serializers/model_serializers.py L518-L534.
Attack Vector
Exploitation requires an authenticated account with permission to register subscription filter keys and create subscriptions within an organization. The attacker submits a filter key containing SQL metacharacters that satisfy the keyword regex, attaches it to a subscription, and triggers any API path that computes usage aggregation. The server builds a GROUP BY clause that includes the attacker's payload verbatim, executing it against the shared database. Further technical analysis is available in the HackMD write-up and the VulnCheck advisory.
No verified public exploit code is published. Readers should consult the references above for proof-of-concept details rather than relying on synthetic payloads.
Detection Methods for CVE-2026-115332
Indicators of Compromise
- Rows in the subscription filter tables where the key column contains SQL metacharacters such as quotes, semicolons, parentheses, or the strings UNION, SELECT, --, or /*.
- Database logs showing GROUP BY clauses that include tokens not present in the application's static schema or allow-list.
- API audit entries where the same organization user registers an unusual filter key and immediately creates or updates a subscription referencing it.
- Unexpected query errors or long-running aggregation queries originating from the Lotus metering_billing service.
Detection Strategies
- Enable PostgreSQL log_statement = 'mod' or all on the Lotus database and alert on identifiers inside GROUP BY that do not match the expected column allow-list.
- Add application-side logging around subscription filter creation that records the raw submitted key and the authenticated user, then review for anomalous values.
- Hunt in historical logs for subscription filter keys that fail to match a strict ^[A-Za-z0-9_]+$ pattern.
Monitoring Recommendations
- Monitor database error rates and slow query counts tied to usage aggregation endpoints for sudden spikes.
- Track per-tenant volumes of newly registered filter keys and flag accounts that register many unique keys in short windows.
- Correlate authentication logs with filter-key creation events to identify compromised or abusive organization accounts.
How to Mitigate CVE-2026-115332
Immediate Actions Required
- Inventory Lotus deployments and confirm the running version against the uselotus/lotus repository; treat any build at or below 0.10.2 as vulnerable.
- Restrict API access to the subscription filter endpoints to trusted operators until a patched build is deployed.
- Audit the subscription filter tables for keys containing non-alphanumeric characters and remove or quarantine suspicious rows.
- Rotate database credentials used by the Lotus backend if cross-tenant access is suspected.
Patch Information
At the time of NVD publication on 2026-10-22, no fixed version is referenced in the advisory data. Track the VulnCheck advisory and the Lotus GitHub repository for an upstream patch. Apply the vendor fix as soon as it ships and redeploy any downstream forks that embed the vulnerable serializer.
Workarounds
- Enforce a strict server-side allow-list on subscription_filter_keys that rejects any value not matching ^[A-Za-z0-9_]{1,64}$ before persistence.
- Patch the serializer in backend/metering_billing/serializers/model_serializers.py to quote identifiers via the database adapter rather than string concatenation.
- Place the Lotus API behind a web application firewall rule that blocks SQL metacharacters in request bodies targeting subscription endpoints.
- Where feasible, isolate tenants into separate databases or schemas to limit blast radius if the serializer is exploited.
# Configuration example: PostgreSQL audit logging for Lotus
ALTER SYSTEM SET log_statement = 'mod';
ALTER SYSTEM SET log_min_duration_statement = 250;
SELECT pg_reload_conf();
# Revoke write access to filter tables from the application role until patched
REVOKE INSERT, UPDATE ON subscription_filter FROM lotus_app;
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

