Back to skill

Security audit

Huo15 Comic Storyboard

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its storyboard purpose, but a crafted character manifest could cause local files to be uploaded to the external image API.

Install only if you trust the script.json and character manifest sources. Before use, constrain reference image paths to files inside the character directory, reject symlinks and absolute/traversal paths, validate image type and size, and treat all prompts and reference images as data sent to Ark/Volcengine using ARK_API_KEY.

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 (1)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/_shared/ark_api.py:225
Finding

Arbitrary Local File Disclosure Through Unvalidated Reference-Image Paths

Content
View full analysis
list[str]: refs = [] for cid in char_ids[:MAX_REF]: info = char_manifest.get(cid, {}) for img in info.get("images", []): if "_full." in img: refs.append(img) break return refs ``` From `scripts/storyboard.py:79-80`: ```python prompt = build_prompt(style, scene) refs = pick_refs(scene.get("characters", []), char_manifest) ``` From `scripts/_shared/ark_api.py:54-59`: ```python content = [{"type": "text", "text": prompt}] for img in reference_images or []: content.append({ "type": "image_url", "image_url": {"url": self._image_to_data_uri(img)}, }) ``` From `scripts/_shared/ark_api.py:225-235`: ```python @staticmethod def _image_to_data_uri(path_or_url: str) -> str: if path_or_url.startswith(("http://", "https://", "data:")): return path_or_url p = pathlib.Path(path_or_url) ext = p.suffix.lower().lstrip(".") or "jpeg" if ext == "jpg": ext = "jpeg" b64 = base64.b64encode(p.read_bytes()).decode() return f"data:image/{ext};base64,{b64}" ``` ### Technical Analysis The character manifest is loaded from the user-selected character directory, and its `images` entries are treated as trusted paths. `pick_refs()` only checks whether the supplied string contains `_full.`. It does not ensure that the path: - Is relative to the specified character directory. - Resolves to a file beneath that directory. - Is a regular, non-symlink image file. - Has an approved image extension and valid image signature. - Is within an acceptable size limit. The accepted path is passed to `_image_to_ ...[truncated 2224 chars]
Remediation
View remediation
pathlib.Path: root = root.resolve(strict=True) candidate = pathlib.Path(value) if candidate.is_absolute(): raise ValueError("Absolute reference-image paths are not allowed") candidate = (root / candidate).resolve(strict=True) try: candidate.relative_to(root) except ValueError as exc: raise ValueError("Reference image escapes the character directory") from exc if candidate.is_symlink() or not candidate.is_file(): raise ValueError("Reference image must be a regular, non-symlink file") if candidate.suffix.lower() not in {".png", ".jpg", ".jpeg", ".webp"}: raise ValueError("Unsupported reference-image format") if candidate.stat().st_size > 20 * 1024 * 1024: raise ValueError("Reference image exceeds the size limit") return candidate ``` Image decoding should subsequently verify that the content i ...[truncated 62 chars]
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (21)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Budget enforcement, cost persistence, and downgrade logic are not inherently unsafe, but they materially exceed the narrow storyboard behavior declared in the skill description. Hidden cost/state management can persist execution metadata to disk and indicates the skill is part of a larger pipeline than advertised, reducing transparency and increasing the likelihood of misuse or unintended side effects.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Budget enforcement, cost persistence, and downgrade logic are not inherently unsafe, but they materially exceed the narrow storyboard behavior declared in the skill description. Hidden cost/state management can persist execution metadata to disk and indicates the skill is part of a larger pipeline than advertised, reducing transparency and increasing the likelihood of misuse or unintended side effects.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Budget enforcement, cost persistence, and downgrade logic are not inherently unsafe, but they materially exceed the narrow storyboard behavior declared in the skill description. Hidden cost/state management can persist execution metadata to disk and indicates the skill is part of a larger pipeline than advertised, reducing transparency and increasing the likelihood of misuse or unintended side effects.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

Budget enforcement, cost persistence, and downgrade logic are not inherently unsafe, but they materially exceed the narrow storyboard behavior declared in the skill description. Hidden cost/state management can persist execution metadata to disk and indicates the skill is part of a larger pipeline than advertised, reducing transparency and increasing the likelihood of misuse or unintended side effects.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Budget enforcement, cost persistence, and downgrade logic are not inherently unsafe, but they materially exceed the narrow storyboard behavior declared in the skill description. Hidden cost/state management can persist execution metadata to disk and indicates the skill is part of a larger pipeline than advertised, reducing transparency and increasing the likelihood of misuse or unintended side effects.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill declares executable behavior that implies filesystem access, environment access, and network use, but it does not declare any explicit tool scope or permissions. This creates a trust gap: reviewers and runtime policy cannot easily constrain what the skill is allowed to do, increasing the risk of overprivileged execution or undisclosed data access.

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
87% confidence
Finding

This code sends prompt text and potentially user-supplied local images, reference media, or audio to remote Volcengine/Ark endpoints via HTTP requests. While the module docstring identifies the API wrapper, there is no visible confirmation prompt or user-facing disclosure near these transmissions, which is relevant because user content and system files may be uploaded externally.

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 script.json and character cards to generate one storyboard keyframe image per shot using Seedream image generation. This file also exposes full video generation workflow methods and text-to-speech synthesis, which are materially broader media-generation capabilities than the stated storyboard-keyframe purpose.

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,

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 ------------

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
95% confidence
Finding

User-facing strings and prompt construction are hard-coded in Chinese, including the module description, status messages, and generated prompt text. This imposes a specific language/locale behavior without any visible opt-in or configuration, matching the language-policy violation criterion.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

Access to sensitive environment variables is a safety-relevant operation under this policy. Although the code raises an error when the key is missing, it does not visibly disclose to the user that the skill depends on and will access the ARK_API_KEY credential.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

This function creates directories and writes downloaded data to disk, which is a file-write operation covered by the warning requirement. The function name implies downloading, but there is no explicit comment, warning, or user-facing message about saving remote content to the specified local path.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

This code posts user text to a remote speech endpoint and writes the returned audio bytes to disk. Although the docstring says it writes to out_path, there is no explicit safety-oriented disclosure or user-visible notice about transmitting text externally and creating or overwriting a local audio file.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

The only user-facing natural-language instruction in this file is written in Chinese, indicating a language-specific behavior without any visible opt-in or documented language choice. Under the policy, forcing a specific language or locale without user choice is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

This Python file contains natural-language comments, docstrings, and user-facing labels exclusively in Chinese, which can amount to a language/locale policy issue when no user opt-in or justification is provided. The stated policy says to flag skills that force a specific language without user choice, and no alternative language option or region-specific rationale appears in this file.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

This code file contains user-facing natural-language documentation entirely in Chinese, with no indication that language selection is configurable or intentionally limited to a specific locale. Under the policy rule, forcing a specific language without opt-in can be a locale-policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

This code performs an HTTP GET to retrieve image data and saves the response to a local file, which are safety-relevant operations under the warning criteria for code files. While there are progress prints elsewhere, there is no explicit disclosure here or in comments/docstrings warning that external network access and filesystem writes will occur.

Content

No source excerpt is available for this finding.