Back to skill

Security audit

translate-book

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because it reports a risk of copying unrelated local files into translation outputs and flags installation or release commands that run unpinned packages; the scan was partial.

Consider the reported risk that a crafted book could cause unrelated local files to be included in translation outputs, and the fact that installation and publishing examples use packages without fixed versions. Install only if you trust the source and are comfortable reviewing those commands and keeping sensitive files away from the workflow. The scan covered 97.1% of components and was marked partial; some text checks were degraded, two images were out of scope, and some references could not be resolved, so the evidence is incomplete.

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

Warning
Location
scripts/convert.py
Finding

Book-controlled image paths can copy files outside the selected source directory

Content
View full analysis

Impact

An attacker-authored book can cause readable local secrets or other unrelated files to be duplicated into the translation artifacts, without image-content validation or additional approval. Disclosure to another party would additionally require those artifacts to be shared or otherwise exposed; no automatic network exfiltration is established.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Rogue AgentSelf-Modification, Session Persistence
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The supplied command performs skill installation, not document conversion or translation. Although the installed skill's name matches the declared purpose, its contents and translation behavior are not provided. Global installation changes the agent environment and is not part of the declared translation workflow.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The supplied command performs software installation and modifies the global skill configuration rather than translating a book. Although it references a translation-related skill and appears in installation documentation, that installation behavior is not disclosed in the declared purpose. The installed skill's translation capabilities cannot be verified from this chunk.

Content

No source excerpt is available for this finding.

Code capability 'network' detected in scripts/convert.py but not covered by declared permissions.

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'network' capability that is not covered by the declared 'allowed-tools' preapproval. Review the host runtime's actual enforcement policy; runtime permissions are unknown unless separately verified, and this declaration alone does not establish a permission ceiling.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx clawhub'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The release instructions execute clawhub using the mutable @latest tag rather than a reviewed, fixed version. Although this is a publishing CLI, not an MCP server, a compromised future package could execute with the maintainer's permissions and potentially access publishing credentials or alter releases; this exposure occurs during releases, not ordinary translation.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx clawhub'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The documented release command runs clawhub@latest, allowing the executable package version to change without review. This is not an MCP configuration, but it is a supply-chain weakness: a compromised package could run with release-account permissions and access credentials or modify published artifacts.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The Codex installation command invokes an unversioned npm executable through npx, which can download and execute a changing package release. It is an installer rather than an MCP server, but a compromised release could execute with the user's permissions; the global installation context also makes reviewing the installed skill important.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 74)May include surrounding context.

Or install it manually:

bash
mkdir -p ~/.agents/skills
git clone https://github.com/deusyu/translate-book.git ~/.agents/skills/translate-book

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.zh-CN.md (reported line 74)May include surrounding context.

或手动安装:

bash
mkdir -p ~/.agents/skills
git clone https://github.com/deusyu/translate-book.git ~/.agents/skills/translate-book

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The Claude Code installation example executes skills through npx without fixing the package version. Although no MCP server is involved, a compromised package selected at installation time could run arbitrary code with the user's privileges; this is an installation-time supply-chain risk, not evidence of malicious translation behavior.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 106)May include surrounding context.

In the Codex CLI or IDE extension, enter:

text
$translate-book Translate /path/to/book.pdf into Chinese.

Codex can also select the skill automatically when your request matches its description.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.zh-CN.md (reported line 106)May include surrounding context.

在 Codex CLI 或 IDE 扩展中输入:

text
$translate-book Translate /path/to/book.pdf into Chinese.

当请求与 Skill 描述匹配时,Codex 也可以自动选择该 Skill。

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · README.md (reported line 358)May include surrounding context.

md
If you find this project helpful, please consider giving it a Star ⭐!

[![Star History Chart](https://api.star-history.com/svg?repos=deusyu%2Ftranslate-book&type=Date%29%5D%28https%3A%2F%2Fstar-history.com%2F)

## Sponsor

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 · README.zh-CN.md (reported line 358)May include surrounding context.

md
如果这个项目对您有帮助,请考虑为其点亮一颗 Star ⭐!

[![Star History Chart](https://api.star-history.com/svg?repos=deusyu%2Ftranslate-book&type=Date%29%5D%28https%3A%2F%2Fstar-history.com%2F)

## 赞助

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The Codex installation example runs an unversioned skills package through npx, allowing the executed release to change over time. This is not an MCP server reference, but a compromised installer release could execute with the user's permissions before installing the skill.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The Claude Code installation command does not pin the npm package executed by npx. Although the scanner incorrectly calls it an MCP server, the mutable executable dependency creates an installation-time supply-chain risk if a future release is compromised.

Content

No source excerpt is available for this finding.

Book content is passed to translation sub-agents without a privacy warning about model/API processing.

Medium
Category
Output Handling
Confidence
86% confidence
Finding

The workflow instructs the agent to launch API-rate-limited sub-agents and supply source chunks, glossary terms, and neighboring excerpts. It does not warn users that, when the runtime uses a hosted model, these contents will be transmitted to its model provider, which matters for confidential or unpublished books.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
85% confidence
Finding

The poster displays an installation command that invokes an unversioned npm executable through npx. The layout asset does not execute it itself, reducing direct exposure, but users copying the advertised command could execute a compromised future installer release with their own permissions; it is not an MCP server configuration.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/calibre_html_publish.py (reported line 28)May include surrounding context.

python
for path in possible_paths:
        try:
            result = subprocess.run([path, "--version"], capture_output=True, text=True, timeout=10)
            if result.returncode == 0:
                print(f"Found Calibre ebook-convert: {path}")
                return path

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/convert.py (reported line 42)May include surrounding context.

python
for path in possible_paths:
        try:
            result = subprocess.run([path, "--version"], capture_output=True, text=True, timeout=10)
            if result.returncode == 0:
                print(f"Found Calibre ebook-convert: {path}")
                return path

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/calibre_html_publish.py (reported line 260)May include surrounding context.

python
try:
        print(f"Starting conversion (timeout: {timeout}s)...")
        result = subprocess.run(cmd, capture_output=True, text=True, timeout=timeout)
        
        if result.returncode == 0:
            if os.path.exists(output_file):

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/convert.py (reported line 57)May include surrounding context.

python
try:
        print(f"Converting {input_file} to HTMLZ...")
        cmd = [calibre_path, input_file, htmlz_file]
        result = subprocess.run(cmd, capture_output=True, text=True, timeout=600)

        if result.returncode == 0:
            file_size = os.path.getsize(htmlz_file)

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/merge_and_build.py (reported line 480)May include surrounding context.

python
def check_pandoc_available():
    """Check if pandoc is available"""
    try:
        subprocess.run(['pandoc', '--version'], capture_output=True, check=True)
        return True
    except (subprocess.CalledProcessError, FileNotFoundError):
        return False

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/merge_and_build.py (reported line 501)May include surrounding context.

python
]

    try:
        subprocess.run(cmd, capture_output=True, text=True, check=True)
        print("Converted with pandoc")
        return True
    except subprocess.CalledProcessError as e:

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/merge_and_build.py (reported line 988)May include surrounding context.

python
cmd = ["python3", publish_script, html_file, "-o", output_file, "--lang", lang_attr]
        if cover:
            cmd.extend(["--cover", cover])
        result = subprocess.run(cmd, check=True, capture_output=True, text=True)

        if os.path.exists(output_file):
            file_size = os.path.getsize(output_file)