Back to skill

Security audit

AI Stem Splitter

Security checks for vulnerabilities and agentic risk

Overview

This skill coherently helps split audio through a hosted third-party API, with expected but privacy-relevant upload and dependency considerations.

Install only if you are comfortable sending the chosen audio or source URL to AI Stem Splitter and using an API key that may consume credits. For sensitive projects, prefer direct REST calls or pinned SDK versions in an isolated environment, and avoid uploading audio you do not have rights or permission to process.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
references/api.md:65
Finding
Unpinned Third-Party SDK Dependencies## Vulnerability Details **File Location**: `references/api.md`, lines 65-75 **Vulnerability Type**: `T08: Insecure Dependencies` **Risk Level**: Medium **Complete Code Snippet**: ```markdown ## SDKs Node: ```bash npm install @aistemsplitter/sdk ``` Python: ```bash pip install aistemsplitter ``` ``` ### Technical Analysis The documented installation commands retrieve the latest available versions of third-party packages without an exact version, lockfile, integrity hash, or documented provenance verification. Consequently, the code installed during different executions can change without any corresponding change to the reviewed Skill. This creates a supply-chain risk if either registry package is compromised, transferred to an untrusted maintainer, subject to account takeover, or updated with malicious code. Package-controlled installation hooks or code executed when the SDK is imported could then run under the privileges of the user or agent environment that followed these instructions. The project contains no evidence that these packages are currently malicious. The vulnerability is the unbounded trust placed in mutable third-party releases. ### Attack Path 1. An attacker compromises the publisher account or release process for `@aistemsplitter/sdk` or `aistemsplitter`. 2. The attacker publishes a malicious package version to the relevant package registry. 3. An agent or user follows the documented unpinned `npm install` or `pip install` command. 4. The package manager resolves the command to the attacker-controlled release. 5. Malicious code executes through an installation lifecycle hook or when the SDK is imported and used. 6. The payload operates with the privileges and data access of the installing process. ### Impact Assessment Successful exploitation could allow arbitrary code execution within the installation or runtime environment. Depending on that environment's permissions, an att ...[truncated 446 chars]
Remediation
## Remediation Suggestions 1. Pin each SDK to a reviewed exact version rather than resolving the latest release. 2. Commit ecosystem-appropriate lockfiles and enforce integrity verification during installation. 3. Record and verify the packages' official publisher identities and source repositories. 4. Review package contents, dependency trees, and installation lifecycle scripts before approving upgrades. 5. Use automated dependency scanning and controlled update workflows. 6. Install dependencies in an isolated, least-privileged environment without unrelated credentials or filesystem access. 7. Prefer direct HTTPS REST requests when SDK functionality is unnecessary, reducing the third-party dependency surface. 8. If example commands must remain generic, explicitly warn users to substitute an approved pinned version and verify package integrity.
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 (8)

External Transmission

Medium
Category
Data Exfiltration
Content
```bash
export BASE_URL="https://api.aistemsplitter.org"

curl -sS "$BASE_URL/v1/credits" \
  -H "Authorization: Bearer $AISTEMSPLITTER_API_KEY"
```
Confidence
60% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
- Product name: AI Stem Splitter
- Core output: vocals, drums, bass, guitar, piano, other
- Common use cases: karaoke vocal removal, acapella extraction, DJ remix prep, guitar or piano practice loops, stem-level audio analysis
- Public API base: `https://api.aistemsplitter.org/v1`
- API key source: AI Stem Splitter Settings -> Developer
- Authentication: `Authorization: Bearer $AISTEMSPLITTER_API_KEY`
- Credits: one credit equals one second of source audio
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
- Product name: AI Stem Splitter
- Core output: vocals, drums, bass, guitar, piano, other
- Common use cases: karaoke vocal removal, acapella extraction, DJ remix prep, guitar or piano practice loops, stem-level audio analysis
- Public API base: `https://api.aistemsplitter.org/v1`
- API key source: AI Stem Splitter Settings -> Developer
- Authentication: `Authorization: Bearer $AISTEMSPLITTER_API_KEY`
- Credits: one credit equals one second of source audio
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
- Product name: AI Stem Splitter
- Core output: vocals, drums, bass, guitar, piano, other
- Common use cases: karaoke vocal removal, acapella extraction, DJ remix prep, guitar or piano practice loops, stem-level audio analysis
- Public API base: `https://api.aistemsplitter.org/v1`
- API key source: AI Stem Splitter Settings -> Developer
- Authentication: `Authorization: Bearer $AISTEMSPLITTER_API_KEY`
- Credits: one credit equals one second of source audio
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
- Product name: AI Stem Splitter
- Core output: vocals, drums, bass, guitar, piano, other
- Common use cases: karaoke vocal removal, acapella extraction, DJ remix prep, guitar or piano practice loops, stem-level audio analysis
- Public API base: `https://api.aistemsplitter.org/v1`
- API key source: AI Stem Splitter Settings -> Developer
- Authentication: `Authorization: Bearer $AISTEMSPLITTER_API_KEY`
- Credits: one credit equals one second of source audio
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
- Product name: AI Stem Splitter
- Core output: vocals, drums, bass, guitar, piano, other
- Common use cases: karaoke vocal removal, acapella extraction, DJ remix prep, guitar or piano practice loops, stem-level audio analysis
- Public API base: `https://api.aistemsplitter.org/v1`
- API key source: AI Stem Splitter Settings -> Developer
- Authentication: `Authorization: Bearer $AISTEMSPLITTER_API_KEY`
- Credits: one credit equals one second of source audio
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Check credits:

```bash
curl -sS "https://api.aistemsplitter.org/v1/credits" \
  -H "Authorization: Bearer $AISTEMSPLITTER_API_KEY"
```
Confidence
60% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The API reference instructs users to send audio content and optionally receive webhook event data from an external third-party service, but it does not warn that user-provided audio and related metadata leave the local environment. In a skill that may process user media or URLs, this omission can cause unintended data disclosure and weakens informed consent around third-party transmission.

Static analysis

No suspicious patterns detected.