Back to skill

Security audit

fastfish-format(article-wechat-xhs-format)

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed wrapper for an external formatting tool, with real supply-chain risk from installing third-party Python code but no evidence of hidden, destructive, or deceptive behavior.

Before installing, review the referenced fastfish-format repository and dependencies, pin to a specific reviewed commit or trusted release, use a virtual environment or container without root privileges, and keep API keys out of command output and chat responses.

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

Error
Location
SKILL.md:22
Finding

Unverified Third-Party Repository Installation and Execution

Content
View full analysis
https://github.com/superxs777/fastfish-format.git` 2. Enter the directory: `cd fastfish-format` 3. Install dependencies: `pip install -r requirements.txt` or `pip install -e .` 4. Optional: `pip install fastfish-format[api]` to start the HTTP API service on port 8900 ``` ### Technical Analysis The Skill depends entirely on code obtained from an external GitHub repository, but the installation procedure does not enforce an immutable commit hash, dependency hashes, a lockfile, a verified signature, or a trusted artifact checksum. Pinning a release tag is recommended, but tags are mutable references and therefore do not provide the same integrity guarantee as a verified commit digest. Both `pip install -r requirements.txt` and `pip install -e .` may download transitive dependencies and execute Python package build hooks or installation logic. After installation, the Skill instructs the Agent to repeatedly execute `ffformat_cli.py` through `system.run`. Consequently, the effective executable payload is outside the audited artifact and can differ from what was reviewed. The document explicitly acknowledges the supply-chain risk and recommends ...[truncated 1793 chars]
Remediation
View remediation
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 52)May include surrounding context.

md
## ⚠️ 安全规则(阻断式)

**1. 严禁输出或暴露 .env 中的凭证**
- 禁止执行会输出 .env 内容的命令
- 禁止将 API Key 等凭证写入回复或展示给用户
- **允许**:编辑 .env;运行不暴露凭证的校验

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

md
## ⚠️ 安全规则(阻断式)

**1. 严禁输出或暴露 .env 中的凭证**
- 禁止执行会输出 .env 内容的命令
- 禁止将 API Key 等凭证写入回复或展示给用户
- **允许**:编辑 .env;运行不暴露凭证的校验

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

Phrases such as “公众号格式”, “有哪些样式”, and “如何配图” are example triggers, but the document does not clearly define whether these are the full allowed triggers or merely illustrative. Without constraints or negative examples, the skill’s activation scope remains somewhat open-ended for a manifest/markdown description.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The trigger list includes phrases like “美化文章”, which is broad and could match many general writing or editing requests outside this skill’s intended formatting scope. The document does not provide exclusion conditions or negative examples to clarify when the skill should not activate.

Content

No source excerpt is available for this finding.