Back to skill

Security audit

Huo15 Comic Dub

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly performs TTS dubbing as advertised, but it has under-scoped file-writing and extra media API capabilities that warrant careful review before installation.

Install only if you are comfortable sending dialogue text to Volcengine Ark and creating local project sidecar files. Use trusted script.json inputs, keep output directories isolated, and avoid invoking the bundled image/video/download helpers unless they are reviewed and constrained.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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 (2)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/dub.py:57
Finding

Path Traversal Allows File Writes Outside the Intended Output Directory

Content
View full analysis
Remediation
View remediation
str: if not isinstance(value, str) or not SAFE_ID.fullmatch(value): raise ValueError(f"Invalid {field}") return value ``` 2. Resolve the final destination and enforce containment within the resolved output directory: ```python base = out_dir.resolve() candidate = (base / f"{sid}_{cid}_{idx}.wav").resolve() if candidate != base and base not in candidate.parents: raise ValueError("Output path escapes the output directory") ``` 3. Reject absolute paths, `..`, path separators, null bytes, and identifiers of unexpected types. 4. Consider generating filenames from trusted internal identifiers rather than raw input values. If source identifiers must be retained, map them to sanitized names or stable hashes. 5. Avoid overwriting existing files by using exclusive creation where practical, or require an explicit overwrite option. ]]>

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/_shared/ark_api.py:157
Finding

Unrestricted Video Downloader Enables SSRF and Arbitrary File Overwrite

Content
View full analysis
pathlib.Path: out_path = pathlib.Path(out_path) out_path.parent.mkdir(parents=True, exist_ok=True) r = requests.get(url, stream=True, timeout=300) with open(out_path, "wb") as f: for chunk in r.iter_content(8192): f.write(chunk) return out_path ``` ### Technical Analysis The `download_video()` method accepts a caller-controlled URL and destination path without applying restrictions to either value. The URL is passed directly to `requests.get()`. Consequently, a caller can request loopback, private-network, link-local, or cloud metadata endpoints. Redirects are followed by default, and the method does not revalidate the destination after a redirect or DNS resolution. The output path is also unrestricted. The method creates parent directories and opens the selected destination in write mode, truncating an existing writable file. No containment check limits writes to a dedicated media directory. Additional hardening controls are absent: - No URL scheme or hostname allowlist - No private, loopback, link-local, or reserved-address rejection - No redirect validation - No response-size limit - No HTTP status validation through `raise_for_status()` - No content-type validation - No exclusive file creation or output-directory confinement This method is not invoked by the current `scripts/dub.py` entry point. Therefore, immediate exposure through the documented workflow is limited. However, it is packaged in a TTS-only Skill despite not being required for the declared functionality, and it becomes exploitable if another entry point or caller passes untrusted arguments to it. ### Attack Path 1. An attacker ...[truncated 1315 chars]
Remediation
View remediation
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
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (26)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Cost tracking, budget enforcement, and persistence to .cost.json are not inherently unsafe, but they are materially different from the narrow advertised dubbing behavior and introduce hidden state and side effects. In this context the main risk is unauthorized or confusing file creation, retention of usage metadata, and unexpected operational behavior rather than direct code execution.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Cost tracking, budget enforcement, and persistence to .cost.json are not inherently unsafe, but they are materially different from the narrow advertised dubbing behavior and introduce hidden state and side effects. In this context the main risk is unauthorized or confusing file creation, retention of usage metadata, and unexpected operational behavior rather than direct code execution.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Cost tracking, budget enforcement, and persistence to .cost.json are not inherently unsafe, but they are materially different from the narrow advertised dubbing behavior and introduce hidden state and side effects. In this context the main risk is unauthorized or confusing file creation, retention of usage metadata, and unexpected operational behavior rather than direct code execution.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Cost tracking, budget enforcement, and persistence to .cost.json are not inherently unsafe, but they are materially different from the narrow advertised dubbing behavior and introduce hidden state and side effects. In this context the main risk is unauthorized or confusing file creation, retention of usage metadata, and unexpected operational behavior rather than direct code execution.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Cost tracking, budget enforcement, and persistence to .cost.json are not inherently unsafe, but they are materially different from the narrow advertised dubbing behavior and introduce hidden state and side effects. In this context the main risk is unauthorized or confusing file creation, retention of usage metadata, and unexpected operational behavior rather than direct code execution.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill declares no explicit tool or permission scope even though its described behavior implies file access, network calls to an external TTS provider, and file writes. Missing scope boundaries can let an agent invoke the skill with broader-than-expected capabilities, increasing the chance of unintended data access or external transmission.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

Broad trigger phrases increase the chance that the skill activates in contexts where the user did not intend external TTS processing, file writes, or billable operations. Because this skill handles dialogue content and may call outside services, ambiguous activation raises privacy and cost risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The description omits a clear warning that user dialogue content is transmitted to an external TTS provider and that usage incurs cost. This weakens informed consent and can expose sensitive script content or create unexpected charges, which is especially important for a content-processing skill.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest describes a skill that reads dialogue, assigns voices by character, and generates WAV files via TTS. This module, however, also exposes full image generation, video task submission/polling, and video downloading functionality, which are materially broader capabilities than role-based dubbing and are not required to implement the stated behavior.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

For a skill whose purpose is dialogue dubbing to WAV files, generating images, creating videos, polling long-running video jobs, and downloading remote video outputs are unrelated capabilities. These add materially different media-creation behaviors that are not an obvious implementation detail of TTS dubbing.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Local file paths passed as reference images are automatically read, base64-encoded, and transmitted to an external API, which can expose sensitive local media without any consent or visibility mechanism. In a skill ecosystem, this is dangerous because callers may not realize that a local path causes file exfiltration to a third-party service.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% 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 · scripts/_shared/ark_api.py (reported line 66)May include surrounding context.

python
"size": size,
            "watermark": False,
        }
        resp = requests.post(
            f"{self.base_url}/images/generations",
            headers=self.headers,
            json=body,

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

Reference audio and video values are forwarded to the remote generation API without any user-facing disclosure or validation. This can leak sensitive media or internal URLs to a third-party service, especially in agent workflows where inputs may come indirectly from user files or internal systems.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% 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 · scripts/_shared/ark_api.py (reported line 144)May include surrounding context.

python
if seed is not None:
            body["seed"] = seed

        resp = requests.post(
            f"{self.base_url}/contents/generations/tasks",
            headers=self.headers,
            json=body,

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The TTS method sends arbitrary input text to an external service and writes the returned audio locally, but the wrapper provides no disclosure, consent, or data-sensitivity controls. For a dubbing skill, the external transmission is expected, yet scripts may contain private dialogue or unpublished content that users should be warned about before upload.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% 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 · scripts/_shared/ark_api.py (reported line 207)May include surrounding context.

python
if emotion:
            body["emotion"] = emotion

        resp = requests.post(
            f"{self.base_url}/audio/speech",
            headers=self.headers,
            json=body,

Tainted flow: 'resp' from requests.get (line 158, network input) → pathlib.Path.write_bytes (file write)

Medium
Category
Data Flow
Confidence
65% confidence
Finding

Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.

Content

Scanner excerpt · scripts/_shared/ark_api.py (reported line 216)May include surrounding context.

python
resp.raise_for_status()
        out = pathlib.Path(out_path)
        out.parent.mkdir(parents=True, exist_ok=True)
        out.write_bytes(resp.content)
        return out

    # ------------ 工具 ------------

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The module docstring is written only in Chinese and provides no indication that language is selectable or user-configurable. Under the stated policy, forcing a specific language or locale without opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

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 · scripts/_shared/config.py (reported line 35)May include surrounding context.

python
# API 端点(2026-04 核对)
ENDPOINTS = {
    "ark_base":      "https://ark.cn-beijing.volces.com/api/v3",
    "kling_base":    "https://api.klingai.com/v1",         # 官方;第三方 piapi.ai/fal.ai 亦可
    "suno_base":     "https://api.sunoapi.org",             # 第三方(Suno 无公开官方 API)
}

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This code file contains natural-language docstrings, exception messages, and warnings entirely in Chinese, including the primary module description and runtime user-facing output. The file does not offer a language choice or document a justified locale-specific constraint, which violates the stated language/locale policy criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

Natural-language strings and documentation in this file are entirely in Chinese, including the module description and runtime messages, with no indication that users may opt into another language. Under the language/locale policy, forcing a specific language without user choice or a documented justification is a policy concern.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The code sends each dialogue line (text) to an external TTS provider via client.tts(...), which can expose potentially sensitive script content, character names, or unpublished material to a third party. In this skill's context, transmitting full dialogue is the core function, but the absence of explicit notice, consent, or data-handling controls still creates a real privacy and confidentiality risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

Natural-language documentation in the module is presented in Chinese, and the code also defaults to a Chinese TTS voice. There is no accompanying language-choice or opt-in text, which may violate a language/locale policy requiring user choice rather than forcing a specific language context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The download_video method fetches remote content and writes it to a specified local path, creating parent directories as needed. This is a file-write operation with no confirmation prompt, logging, or warning in the function documentation to disclose that local files will be created or overwritten.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The save method creates the project directory and writes .checkpoint.json containing workflow status to disk, but there is no confirmation prompt, print/log message, or explanatory comment/docstring warning that the skill persists state on the user's filesystem. For a code file, file writes can warrant disclosure when the persistence behavior is not otherwise surfaced to the user.

Content

No source excerpt is available for this finding.