Back to skill

Security audit

vestafolio-rentabilite-locative

Security checks for vulnerabilities and agentic risk

Overview

This skill is a clearly disclosed Vestafolio rental-yield API helper that sends only user-approved simulator inputs, with no persistence or credential use.

Install only if you are comfortable sending the listed property-investment assumptions to Vestafolio after reviewing the exact values. Do not approve the API call for information you consider private unless you are satisfied with Vestafolio's handling of that data.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

other

Warning
Location
SKILL.md:15
Finding

Unnecessary External Transmission of Sensitive Financial Data

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 15-32
Vulnerability Type: Excessive external transmission of financial data
Risk Level: Medium

The Skill requires property and investment-related financial data to be submitted to Vestafolio's external API, even though SKILL.md documents the formulas necessary to perform the calculation locally.

Vulnerable Code Snippet:

text
1. Reuse answers already supplied. Ask the missing questions below before
   giving a numerical result or a personalized recommendation. Example values
   and schema defaults are not the user's answers.
2. Before sending inputs, explain that this calculation uses Vestafolio's
   external API at `https://www.vestafolio.com`, summarize the fields and
   values to be sent, and obtain the user's agreement. Existing explicit
   approval for this destination and data scope is sufficient; do not ask
   again for calls it already covers. Installing or selecting the skill alone
   is not approval to transmit personal data. If the user declines, do not
   POST; offer the simulator link for manual use without prefilled inputs.
3. Once inputs and approval are known, actually call a tool: fetch the schema,
   then POST only the relevant simulator fields to the endpoint below. Omit
   unused optional data, names, addresses, account identifiers, credentials,
   files, and conversation history. No authentication is needed. Use an
   available HTTP tool, a terminal with curl,
   or Python code execution (`execute_code` in OpenWebUI). Python can use
   `urllib.request`; in browser-based Pyodide use `await pyfetch(...)` from
   `pyodide.http`. A Python environment does not need a shell to call the API.

The required request is demonstrated later in the file:

bash
curl --fail-with-body --silent --show-error --max-time 30 -X POST https://www.vestafolio.com/api/tools/v1/rentabilite-locative \
  -H 'Content-Type: appli
...[truncated 3593 chars]
Remediation
View remediation

Remediation Suggestions

  1. Implement the documented formulas locally and make local calculation the default behavior.
  2. Treat the external API as an optional verification mechanism rather than a mandatory dependency.
  3. Before optional API use, continue to identify the destination, enumerate the exact fields and values, and obtain explicit, scope-specific consent.
  4. Preserve the existing data-minimization rules that prohibit sending names, addresses, credentials, account identifiers, files, unused fields, and conversation history.
  5. Publish or link to the service's privacy policy and clearly disclose retention periods, subprocessors, deletion options, and whether submitted data is used for analytics or model training.
  6. Validate all values locally against documented bounds before transmission and send only the minimum fields required by the current schema.
  7. Provide a local fallback when network access is unavailable or consent is declined, instead of refusing to calculate despite having the necessary formulas.
  8. If API parity is operationally important, add tests comparing local calculations with anonymized API test vectors rather than sending each user's real financial inputs.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 115)May include surrounding context.

Then POST the user's parameters (all amounts in euros, rates in percent):

bash
curl --fail-with-body --silent --show-error --max-time 30 -X POST https://www.vestafolio.com/api/tools/v1/rentabilite-locative \
  -H 'Content-Type: application/json' \
  -d '{
    "purchasePrice": 200000,