Audit-Verlauf

Version	Datum	Ergebnis	Prüfelemente	Änderung ggü. vorheriger
v4 Neueste	5. Juli 2026, 17:58	1 bestätigt	12	Keine Änderung der Fähigkeiten
v3	5. Juli 2026, 17:58	1 bestätigt	12	Keine Änderung der Fähigkeiten
v2	30. Juni 2026, 04:22	Keine bestätigten Befunde	2	Keine Änderung der Fähigkeiten
v1	19. Mai 2026, 09:23	Keine bestätigten Befunde	2	Ausgangsbasis

5. Juli 2026, 17:58

Most static backtick alerts are false positives from Markdown formatting, input tables, and sample code blocks. The confirmed issues are the documented belt and npx command flows plus the intentional dependency on hosted inference.sh rendering. No prompt injection text was found in SKILL.md.

1

Gescannte Dateien

213

Analysierte Zeilen

15

Prüfelemente

0

Falschmeldungen ignoriert

Bestätigte Sicherheitsbedenken (1)

Mittel

SKILL.md:11 SKILL.md:23-29 SKILL.md:37-44

Remote Rendering Receives User TSX Code

The skill sends React and Remotion TSX, props, and render settings to inference.sh for hosted rendering. Users could expose proprietary animation code, unreleased content, or sensitive asset references if they submit private project data.

The documentation states rendering is done through inference.sh and shows code submitted through belt app run input. The behavior is intentional, but the data exposure risk is direct.

Elemente der Fähigkeitsprüfung (12)

Dies sind echte lokale Fähigkeiten, die für diese Fähigkeit erwartet werden können; daher müssen sie überprüft werden, werden jedoch nicht als bestätigtes bösartiges Verhalten gezählt.

Mittel

SKILL.md:7

Ruby/shell backtick execution

> **Install the belt CLI skill:** `npx skills add belt-sh/cli`

The instruction asks users to execute npx to add a third-party CLI skill. This is a real supply-chain risk because it installs executable tooling from outside the current package.

Mittel

SKILL.md:17

Ruby/shell backtick execution

> Requires inference.sh CLI (`belt`). [Install instructions](https://raw.githubusercontent.com/infer

The line directs users to install and use the belt CLI before running the skill. The executable command is not embedded on this line, but the dependency creates real external-command exposure.

Mittel

SKILL.md:19-30 SKILL.md:72-80 SKILL.md:84-92 SKILL.md:96-98 SKILL.md:108-117 SKILL.md:121-129 SKILL.md:190-205

Ruby/shell backtick execution · 7 Vorkommen

```bash

The fenced quick-start block contains belt login and belt app run commands. Running those commands executes an external CLI and submits render input to a hosted service.

Niedrig

SKILL.md:11

Hardcoded URL

Render videos from React/Remotion component code via [inference.sh](https://inference.sh) CLI.

The skill explicitly routes rendering through inference.sh. This is disclosed and central to the feature, but it confirms a remote service dependency.

Niedrig

SKILL.md:13

Hardcoded URL

![Remotion Render](https://cloud.inference.sh/app/files/u/4mg21r6ta37mpaz6ktzwtt8krr/01kg2c0egyg243m

The Markdown image uses a hosted cloud.inference.sh asset. Viewing the documentation can trigger a network fetch to a third-party host, which is a low privacy risk.

Niedrig

SKILL.md:17

Hardcoded URL

> Requires inference.sh CLI (`belt`). [Install instructions](https://raw.githubusercontent.com/infer

The install-instructions link points to raw GitHub content for the required CLI. This is an intentional dependency path, but users rely on external network content to set up the tool.

Geprüft von: codex

Audit-Version 3

5. Juli 2026, 17:58

Most static backtick alerts are false positives from Markdown formatting, input tables, and sample code blocks. The confirmed issues are the documented belt and npx command flows plus the intentional dependency on hosted inference.sh rendering. No prompt injection text was found in SKILL.md.

1

Gescannte Dateien

213

Analysierte Zeilen

15

Prüfelemente

0

Falschmeldungen ignoriert

Bestätigte Sicherheitsbedenken (1)

Mittel

SKILL.md:11 SKILL.md:23-29 SKILL.md:37-44

Remote Rendering Receives User TSX Code

The skill sends React and Remotion TSX, props, and render settings to inference.sh for hosted rendering. Users could expose proprietary animation code, unreleased content, or sensitive asset references if they submit private project data.

The documentation states rendering is done through inference.sh and shows code submitted through belt app run input. The behavior is intentional, but the data exposure risk is direct.

Elemente der Fähigkeitsprüfung (12)

Dies sind echte lokale Fähigkeiten, die für diese Fähigkeit erwartet werden können; daher müssen sie überprüft werden, werden jedoch nicht als bestätigtes bösartiges Verhalten gezählt.

Mittel

SKILL.md:7

Ruby/shell backtick execution

> **Install the belt CLI skill:** `npx skills add belt-sh/cli`

The instruction asks users to execute npx to add a third-party CLI skill. This is a real supply-chain risk because it installs executable tooling from outside the current package.

Mittel

SKILL.md:17

Ruby/shell backtick execution

> Requires inference.sh CLI (`belt`). [Install instructions](https://raw.githubusercontent.com/infer

The line directs users to install and use the belt CLI before running the skill. The executable command is not embedded on this line, but the dependency creates real external-command exposure.

Mittel

SKILL.md:19-30 SKILL.md:72-80 SKILL.md:84-92 SKILL.md:96-98 SKILL.md:108-117 SKILL.md:121-129 SKILL.md:190-205

Ruby/shell backtick execution · 7 Vorkommen

```bash

The fenced quick-start block contains belt login and belt app run commands. Running those commands executes an external CLI and submits render input to a hosted service.

Niedrig

SKILL.md:11

Hardcoded URL

Render videos from React/Remotion component code via [inference.sh](https://inference.sh) CLI.

The skill explicitly routes rendering through inference.sh. This is disclosed and central to the feature, but it confirms a remote service dependency.

Niedrig

SKILL.md:13

Hardcoded URL

![Remotion Render](https://cloud.inference.sh/app/files/u/4mg21r6ta37mpaz6ktzwtt8krr/01kg2c0egyg243m

The Markdown image uses a hosted cloud.inference.sh asset. Viewing the documentation can trigger a network fetch to a third-party host, which is a low privacy risk.

Niedrig

SKILL.md:17

Hardcoded URL

> Requires inference.sh CLI (`belt`). [Install instructions](https://raw.githubusercontent.com/infer

The install-instructions link points to raw GitHub content for the required CLI. This is an intentional dependency path, but users rely on external network content to set up the tool.

Geprüft von: codex

Audit-Version 2

30. Juni 2026, 04:22

Static analysis flagged many external command and network patterns. Review found these are mostly documented belt CLI examples and official documentation links, but the skill intentionally sends user-supplied TSX code to inference.sh for remote rendering. No evidence found of prompt injection, hidden scripts, credential theft, or malicious intent in SKILL.md.

1

Gescannte Dateien

213

Analysierte Zeilen

4

Prüfelemente

3

Falschmeldungen ignoriert

Elemente der Fähigkeitsprüfung (2)

Dies sind echte lokale Fähigkeiten, die für diese Fähigkeit erwartet werden können; daher müssen sie überprüft werden, werden jedoch nicht als bestätigtes bösartiges Verhalten gezählt.

Mittel

SKILL.md:11-17 SKILL.md:23-29 SKILL.md:138-169

User Code Sent to Remote Renderer

TRUE POSITIVE for network risk. The skill directs users to render TSX component code through inference.sh using the belt CLI. This is intended, but proprietary code, props, URLs, or assets may be processed by a third-party service.

The documentation explicitly names inference.sh and shows belt app run calls that submit a code field for rendering. This confirms the network exposure, while the context shows it is the stated purpose of the skill.

Mittel

SKILL.md:4 SKILL.md:19-30

Scoped External CLI Execution

TRUE POSITIVE for external command capability. The skill permits Bash(belt *) and provides belt command examples. The scope is limited to the belt CLI, but marketplace users should know the skill can run external commands that contact a remote service.

The allowed-tools declaration is direct evidence of a scoped CLI command surface. I did not find evidence that it executes arbitrary shell commands outside belt.

Statische falsch positive Treffer ignoriert (3)

Diese statischen Treffer wurden durch semantische Prüfung verworfen oder entsprachen reinen Schema-Tokens; daher werden sie aus Transparenzgründen angezeigt, beeinflussen jedoch nicht die Qualitätsbewertung.

Niedrig

SKILL.md:19-30 SKILL.md:72-80 SKILL.md:84-92 SKILL.md:96-104 SKILL.md:108-117 SKILL.md:121-129 SKILL.md:190-205

CLI Example Detections Are Mostly Benign

FALSE POSITIVE for malicious command execution. The detected shell snippets are fenced examples showing how to use the documented belt render workflow and related skill installs. They are not hidden scripts or automatic execution paths.

The command patterns occur inside markdown code fences and match the skill purpose. No separate executable file or stealth command path exists in the reviewed file set.

Niedrig

SKILL.md:11-17 SKILL.md:209-211

Hardcoded URLs Are Documentation and Service References

FALSE POSITIVE for suspicious network destinations. The URLs point to inference.sh, a hosted image, install documentation, and Remotion or inference.sh documentation. No evidence found that secrets are collected or exfiltrated.

The URLs are visible user-facing references, not hidden endpoints. Their placement in descriptions and documentation links supports benign use.

Niedrig

SKILL.md:3 SKILL.md:35

Weak Cryptography Matches Lack Supporting Evidence

FALSE POSITIVE for weak cryptography. The reviewed lines do not show cryptographic functions, hashing, ciphers, or security-sensitive comparisons. No evidence found of MD5, SHA1, DES, or similar weak algorithms in SKILL.md.

The cited locations are metadata prose and a parameter table header, not code using cryptography. Confidence is high because the only scanned file contains no crypto implementation.

Risikofaktoren

⚙️ Externe Befehle (8)

SKILL.md:4 SKILL.md:19-30 SKILL.md:72-80 SKILL.md:84-92 SKILL.md:96-104 SKILL.md:108-117 SKILL.md:121-129 SKILL.md:190-205

🌐 Netzwerkzugriff (4)

SKILL.md:11 SKILL.md:13 SKILL.md:17 SKILL.md:209-211

Erkannte Muster

Bash Tool Allowance for belt CLIRemote Render Invocation Pattern

Geprüft von: codex

Audit-Version 1

19. Mai 2026, 09:23

Static analysis found 41 potential issues. After evaluation: all 3 'weak cryptographic algorithm' findings are false positives ('codec' refers to video codec, not cryptography). All 32 'Ruby/shell backtick execution' findings are false positives (markdown fenced bash code blocks, not Ruby). All 6 'hardcoded URL' findings are false positives (documentation links to legitimate services). The skill is a legitimate Remotion video renderer that uses external commands (belt CLI) and network access (inference.sh API) as documented. No prompt injection or malicious intent detected. Published with medium risk and appropriate warnings.

1

Gescannte Dateien

213

Analysierte Zeilen

4

Prüfelemente

0

Falschmeldungen ignoriert

Elemente der Fähigkeitsprüfung (2)

Dies sind echte lokale Fähigkeiten, die für diese Fähigkeit erwartet werden können; daher müssen sie überprüft werden, werden jedoch nicht als bestätigtes bösartiges Verhalten gezählt.

Mittel

SKILL.md:5 SKILL.md:19-30 SKILL.md:72-80 SKILL.md:190-205

External Command Execution via Bash

The skill declares allowed-tools: Bash(belt *) granting broad shell execution. Multiple bash code examples invoke the belt CLI to send user-provided TSX code to a remote rendering API. While the usage is legitimate, unrestricted bash access poses risk if the AI tool executes commands with modified inputs.

The bash commands are explicit documentation examples for a legitimate video rendering workflow. Risk is medium because the allowed-tools declaration is broad but the actual commands are well-scoped to the belt CLI tool.

Niedrig

SKILL.md:11 SKILL.md:209 SKILL.md:210 SKILL.md:211

Hardcoded URLs to External Services

The skill contains hardcoded URLs to remotion.dev and inference.sh documentation. These are legitimate references for the skill's intended functionality but reveal dependence on external services.

URLs point to official documentation pages for Remotion and inference.sh. No evidence of malicious domains or data exfiltration endpoints.

Risikofaktoren

⚙️ Externe Befehle (10)

SKILL.md:5 SKILL.md:7 SKILL.md:19-30 SKILL.md:72-80 SKILL.md:84-92 SKILL.md:96-104 SKILL.md:108-117 SKILL.md:121-129 SKILL.md:133-170 SKILL.md:190-205

🌐 Netzwerkzugriff (6)

SKILL.md:11 SKILL.md:13 SKILL.md:17 SKILL.md:209 SKILL.md:210 SKILL.md:211

Erkannte Muster

Unrestricted Bash Permission Declaration

Geprüft von: claude

Versionsvergleich

Bestätigte Sicherheitsbedenken (1)

Risikofaktoren

Bestätigte Sicherheitsbedenken (1)

Risikofaktoren

Risikofaktoren

Erkannte Muster

Risikofaktoren

Erkannte Muster