Back to skill

Security audit

Axelhu Local Sdxl

Security checks for vulnerabilities and agentic risk

Overview

This is a mostly local SDXL image-generation skill, with no evidence of malicious behavior, but users should clarify the Feishu delivery claim before using it for sensitive images.

Install only if you intend to use a local ComfyUI/SDXL setup and are comfortable with images being saved locally. Before using sensitive prompts or private images, confirm whether any surrounding agent integration actually sends results through Feishu, because the documentation claims that behavior while the included script does not implement it.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
Findings (13)

Unvalidated Output Injection

High
Category
Output Handling
Confidence
85% confidence
Finding

Model output is used without validation or sanitization. Unvalidated output injected into downstream contexts (SQL, shell, HTML) enables injection attacks and arbitrary code execution.

Content

Scanner excerpt · SKILL.md (reported line 43)May include surrounding context.

python
import subprocess
result = subprocess.run(
    ["python3", "/path/to/scripts/sdxl_generate.py",
     "--prompt", "一只狐狸在森林里",
     "--negative", "模糊, 低质量",

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · references/deployment.md (reported line 30)May include surrounding context.

验证运行

bash
curl http://localhost:8188/system_stats | python3 -c "import sys,json; d=json.load(sys.stdin); print(d['devices'][0]['Name'])"

预期输出:cuda:0 NVIDIA GeForce RTX 3080

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% 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 · SKILL.md (reported line 83)May include surrounding context.

md
"save": {"class_type": "SaveImage",
                 "inputs": {"images": ["decode", 0], "filename_prefix": "sdxl_gen"}}
    }
    r = requests.post(f"{COMFYUI}/prompt", json={"prompt": workflow})
    pid = r.json()["prompt_id"]
    # 等待完成(轮询)
    for _ in range(30):

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

L119 states that after generation the script will save the image and send an image message to the user via Feishu. However, the only concrete code shown in the file either runs a local generation script and prints stdout (L041-L053) or calls the local ComfyUI API and returns a localhost image URL (L057-L091); no Feishu messaging logic appears. This is an active documentation-to-code contradiction about the skill's actual side effects and delivery behavior.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill is presented as suitable for private or sensitive local image generation, but later states that outputs are saved locally and sent via Feishu without a clear up-front warning. That creates a real privacy risk because users may provide sensitive prompts or expect strictly local handling, leading to unintended disclosure through message delivery and local artifact retention.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The argument help at L036 says --output sets the output path, implying the generated file will be written there. In the success path, however, the code assigns final = args.output at L176 but only performs shutil.copy at L180 in the else branch when no --output was supplied, so the documented behavior contradicts the actual implementation and may report a path where no file was created.

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/sdxl_generate.py (reported line 131)May include surrounding context.

python
# 3. 发送请求
    print("[SDXL] 发送生图请求...")
    r = requests.post(f"{COMFYUI_URL}/prompt", json={"prompt": workflow}, timeout=30)
    if r.status_code != 200:
        print(f"ERROR: HTTP {r.status_code}: {r.text}")
        sys.exit(1)

Tainted flow: 'prompt_id' from requests.post (line 141, network input) → requests.get (network output)

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/sdxl_generate.py (reported line 154)May include surrounding context.

python
# 从 history API 查
        for i in range(30):
            time.sleep(2)
            r = requests.get(f"{COMFYUI_URL}/history/{prompt_id}", timeout=10)
            if r.status_code == 200:
                hist = r.json()
                if prompt_id in hist:

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The entire skill documentation, trigger phrases, examples, and invocation guidance are written only in Chinese, with no indication that other languages are supported or that the user can choose their preferred language. This can constitute a language/locale policy issue when a skill implicitly constrains interaction language without opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

The manifest description is entirely in Chinese ("本地 SDXL 生图技能 — ComfyUI + RTX 3080") and provides no indication that language choice is optional or that the skill is intended only for a Chinese-speaking or region-specific audience. Per the policy, locale or language constraints should be opt-in or clearly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

This markdown file is natural-language documentation, and all headings and instructions are presented in Chinese. The policy requires flagging language or locale constraints when the skill forces a specific language without user opt-in, and no alternative language option or justification is provided here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The module docstring, CLI description, help text, and runtime messages are all presented in Chinese. This imposes a language choice on users without opt-in or any indication that the tool is intentionally limited to a Chinese-speaking context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

This shell script includes natural-language comments and usage guidance only in Chinese, which imposes a specific language on users without offering an alternative or explaining a locale-specific requirement. The policy allows language constraints only when the user is given a choice or the constraint is clearly justified.

Content

No source excerpt is available for this finding.