accounted-api

Warn

Audited by Snyk on Nov 19, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (medium risk: 0.30). The required workflow exposes the Accounted inbox: GET /companies/{companyId}/inbox-items and GET .../inbox-items/{id} return the "full reading and e-mail text" of items that outsiders can deliver by e-mail into the Inkorg, so external e-mail bodies reach the agent without it first selecting a specific item.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

  • Direct money access detected (medium risk: 0.30). Direct operation: the skill exposes payment-instruction routes. POST /payment-orders prepares draft bank payments for supplier invoices and POST /payment-orders/approve attests them "so they can be signed" (SKILL.md:437-440). POST /supplier-payment-batches creates a betalfil (pain.001 or LB file) for downstream bank execution (SKILL.md:470), and salary-run payment-file generation is documented at SKILL.md:704. These are financial instructions for downstream execution, not bookkeeping-only or illustrative code, so the exclusions do not apply.

Control: the payment routes are gated by service-enforced access control. Every request requires a bearer API key, and scopes are checked by the service, with a missing scope returning 403 (SKILL.md:36-54). The payment-order and batch routes require suppliers:write or payroll:write. Approval of staged writes is gated by the separate pending_operations:approve scope (SKILL.md:153-155). Payment signing also happens through BankID by a person in the Betalningar page, which is external approval (SKILL.md:442, 449). The confirmed: true flag on high-risk approvals is an agent-followed confirmation and is not counted as a control. Because the operation is direct and has documented service-enforced authentication and external signing, the severity is medium, not high.

Indirect routes such as invoices and journal entries do not change the outcome, since the highest applicable tier across routes governs.

Issues (2)

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W009
MEDIUM

Direct money access capability detected (payment gateways, crypto, banking).

Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 19, 2026, 06:13 PM
Issues
2