/

CVE-2026-48169

Severity:

HIGH RISK

(8.8)

Broken Access Control (CWE-639) in PraisonAI Platform API

CVE-2026-48169

Deferred

Vulnerability type

Category:

Broken Access Control

CWE:

CWE-639, CWE-862

Impact:

Cross-workspace data exposure, unauthorized resource modification, and workspace takeover

Severity:

HIGH RISK

(8.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

PraisonAI Platform API versions before 0.1.4 contain authorization failures that can break workspace isolation. An authenticated workspace member may access, modify, or delete issues and projects in other workspaces by altering resource identifiers, and may also elevate their role or remove other members when member-management endpoints do not enforce role hierarchy.

Affected context

The vulnerability occurs in PraisonAI Platform API deployments where service-layer operations look up issues and projects by primary key without constraining those queries to the requested workspace. It also affects member-management routes whose access checks confirm workspace membership but do not require an appropriately privileged role. The route layer may correctly verify membership, while downstream service logic fails to apply workspace ownership and role boundaries.

Exploit preconditions

  • Attack surface: Authenticated PraisonAI Platform API endpoints for issues, projects, and workspace members
  • Required access: Membership in at least one workspace and a valid authenticated session
  • Trigger point: Requests that supply attacker-selected resource identifiers or invoke member-management operations
  • Why exploitation is possible: Service-layer lookups do not consistently constrain resources by workspace, and member operations accept the minimum member role instead of enforcing an administrator or owner role

Why this happens

The primary issue is an object-level authorization failure mapped to CWE-639: Authorization Bypass Through User-Controlled Key. Resource queries trust a client-supplied identifier without combining it with the caller's authorized workspace scope, allowing identifiers from another workspace to resolve successfully. A related enforcement failure maps to CWE-862: Missing Authorization, because membership checks do not verify whether the caller has sufficient privilege for role changes or member removal. The insecure pattern is separating authentication and membership validation from resource ownership and role validation, then failing to carry the authorization context into service-layer operations.

Potential impact

  • An authenticated member may read data belonging to other workspaces
  • An attacker could modify or delete issues and projects outside their authorized workspace
  • A member may elevate their own workspace role
  • An attacker could remove legitimate owners or administrators
  • Workspace confidentiality, integrity, and availability may be compromised
  • Cross-workspace data manipulation could affect downstream operational decisions

General mitigation guidance

Mitigation typically focuses on enforcing authorization at the service and data-access layers rather than relying only on route-level membership checks.

Common defensive patterns include:

  • Apply workspace ownership predicates to every issue and project lookup, update, and delete operation
  • Bind resource identifiers to the authenticated caller's authorized workspace context
  • Enforce role hierarchy checks for adding members, changing roles, and removing members
  • Use deny-by-default authorization policies for sensitive workspace operations
  • Avoid trusting workspace identifiers supplied solely by clients when the server can derive scope from authenticated context
  • Apply least privilege to workspace roles and protect ownership-transfer operations with explicit controls
  • Add authorization tests covering cross-workspace identifiers and every supported role combination

How teams detect issues like this

Organizations can detect these authorization weaknesses through:

  • Static analysis and code review checks for primary-key lookups that lack workspace or tenant predicates
  • Review signals such as service methods accepting resource IDs without an authorization context
  • Policy checks requiring ownership or tenant filters on read, update, and delete queries
  • Tests that replay valid requests with identifiers from another workspace
  • Role-matrix tests covering member, administrator, and owner behavior for every management endpoint
  • Audit-log monitoring for unusual cross-workspace access, unexpected role elevation, or owner removal
  • Automated security review tools that identify missing authorization checks and inconsistent scope propagation

Sources

SourceLink
github.comgithub.com/MervinPraison/PraisonAI/security/advisories/GHSA-gv23-xrm3-8c62
github.comgithub.com/pypa/advisory-database/tree/main/vulns/praisonai-platform/PYSEC-2026-2935.yaml

Frequently asked

questions

Yes, an authenticated API user may be able to exploit it through network requests if the affected endpoints are reachable and the user belongs to at least one workspace.

No. A low-privilege workspace member may access improperly scoped resources and may perform member-management actions when those actions only require ordinary membership.

Enforce workspace ownership and role authorization in the service and data-access layers for every operation, using server-derived authorization context rather than trusting client-supplied scope.

Use centralized, deny-by-default authorization policies; require tenant predicates on resource queries; test every endpoint across workspace and role boundaries; and monitor audit logs for anomalous access or privilege changes.

Find out what's already
exploitable in your codebase.