Security audit

Stripe API Integration

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed Stripe operations skill with high-impact payment examples, but its local storage and Stripe access are coherent with its purpose and guarded by explicit confirmation rules.

Install only if you want your agent to help with real Stripe operations. Keep the default confirm-each live mode, verify test vs live keys before any command, require explicit approval for refunds, payouts, cancellations, invoice finalization, deletes, and webhook changes, and set your own retention/privacy rules for saved incidents, disputes, evidence packets, and recovery outreach data.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (16)

Context-Inappropriate Capability

Medium

Confidence: 87% confidence
Finding: The instruction to log platform-absorbed seller losses in a separate incidents store broadens the skill from Stripe integration guidance into organizational recordkeeping. While not inherently malicious, it can cause unauthorized persistence of financially sensitive incident data and normalize cross-system writes not required for the immediate payment task.

Context-Inappropriate Capability

Low

Confidence: 87% confidence
Finding: The instruction to log platform-absorbed seller losses in a separate incidents store broadens the skill from Stripe integration guidance into organizational recordkeeping. While not inherently malicious, it can cause unauthorized persistence of financially sensitive incident data and normalize cross-system writes not required for the immediate payment task.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The file instructs the agent to persist incident details, customer-impacting events, and user-stated policy declarations into local files outside the immediate Stripe API task. That creates a data-governance risk because sensitive financial incident information and operational policies may be stored without explicit user consent, minimization, or scope controls, increasing the chance of unintended retention or later exposure.

Context-Inappropriate Capability

Medium

Confidence: 96% confidence
Finding: The skill explicitly instructs the agent to modify files outside its Stripe integration scope, including `~/Clawic/data/finances/accounts.md` and `~/Clawic/data/finances/subscriptions.md`. Cross-skill writes break domain boundaries and can let a Stripe-focused workflow overwrite or contaminate finance records, increasing the risk of unauthorized state changes, bad bookkeeping, or privilege creep if the agent has file-write capability.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The recovery section explicitly encourages using customer email addresses from expired checkout sessions for abandoned-cart outreach, but it omits any warning to verify consent, privacy disclosures, retention limits, or applicable marketing rules. In a payments context this is sensitive because checkout data often includes personal information collected for transaction processing, and repurposing it for email campaigns without notice can create privacy, compliance, and trust risks.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The documentation includes a destructive account deletion operation without an explicit warning about permanence, production impact, or identity/authorization checks. In an agent setting, this raises the risk of accidental deletion of connected accounts or unauthorized privacy-impacting changes if the command is followed blindly.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: This file contains live Stripe API examples that create, update, and delete customers, products, prices, coupons, and promotion codes, but it does not prominently warn that these commands modify production data if used with a live secret key. In a troubleshooting skill, operators may copy-paste commands directly, so the absence of explicit safety guidance increases the risk of accidental destructive or billing-impacting actions.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The file instructs operators to write dispute details and save winning evidence packets locally, which can include customer identifiers, access logs, emails, and other dispute artifacts. Although the document briefly says to redact before uploading, it does not establish concrete data-handling controls for what is stored in local markdown files and artifacts, creating a real risk of retaining unnecessary sensitive customer data in insecure or overbroad locations.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: This section provides copy-pastable commands for irreversible or high-consequence billing actions such as finalizing, sending, paying, voiding, and marking invoices uncollectible, but the execution examples themselves do not repeat an explicit safety warning or confirmation step. In a Stripe operations skill, these actions are especially sensitive because they can create legal/accounting records, charge customers, or alter receivables, so an operator may run them too quickly from context without noticing the earlier prose warning.

Missing User Warnings

Low

Confidence: 87% confidence
Finding: The portal session example creates a direct access link to billing-management functions, including invoice history and payment method changes, without warning that the resulting URL is sensitive and should only be generated for and delivered to the authenticated customer. While this is standard Stripe functionality, omitting the privacy/access-control warning increases the chance that an implementer exposes or mishandles a session URL that grants account-level billing access.

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: The payout example shows how to create a payout but does not prominently warn that this action moves real funds out of the Stripe balance and may be irreversible or operationally sensitive. In the context of a troubleshooting skill, that omission can nudge an agent toward executing a money-moving action during diagnosis rather than using read-only inspection first.

Missing User Warnings

Medium

Confidence: 81% confidence
Finding: Although the text says 'Accept the dispute and stop working on it,' it does not clearly emphasize the business consequence that closing a dispute is effectively a concede/forfeit action with little or no opportunity to reverse. In a support or debugging context, insufficient warning around a final dispute action can lead to accidental loss of recoverable funds.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill provides a one-line immediate deletion command for a live subscription without an adjacent warning that it stops service and can be difficult to reverse operationally. In a Stripe troubleshooting skill, operators may copy commands directly; this increases the chance of accidental destructive action against production subscriptions and customer access.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The pause collection examples expose behaviors like `mark_uncollectible` and `void` but do not clearly warn that they change invoice collectability and accounting state. In this skill context, users are likely to apply examples during incidents, so missing warnings can lead to unintended revenue loss, bad receivable records, or compliance/accounting issues.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: Resetting `billing_cycle_anchor` to `now` can trigger an immediate invoice or charge, but the examples do not place a clear warning next to the command. Because this skill is specifically for fixing billing behavior, an operator may use the snippet to 'correct' dates and unintentionally bill customers instantly, causing disputes and trust damage.

Missing User Warnings

Medium

Confidence: 77% confidence
Finding: The file includes live webhook disable and delete commands without an adjacent safety warning, environment scoping note, or confirmation guidance. In an operational troubleshooting skill for Stripe, this can lead users to disable or remove production endpoints during debugging, causing missed payment, subscription, dispute, or payout events and potentially prolonged business-impacting outages.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical

Code: suspicious.exposed_secret_literal
Location: memory-template.md:62