Back to skill

Security audit

wisely-read-ppt

Security checks for vulnerabilities and agentic risk

Overview

The skill receives a suspicious verdict because a scanner found a risk that image removal could delete text along with images, and the scan was partial.

Review the text-preservation risk and the skill's required tool access before installing. The scan covered all listed components but is marked partial because several local references could not be resolved; an additional scanner completed without reporting findings, but the available evidence does not establish that the skill is safe.

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
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Dynamic import via __import__()

Medium
Category
Dangerous Code Execution
Confidence
75% confidence
Finding

Dynamic import() can load arbitrary modules at runtime, bypassing static analysis and potentially importing malicious code.

Content

Scanner excerpt · scripts/parse.py (reported line 143)May include surrounding context.

python
("pdfplumber", "pdfplumber"), ("docx", "docx"),
                     ("PIL", "PIL")]:
        try:
            __import__(mod); env[key] = True
        except Exception:
            pass
    return env

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/parse.py (reported line 226)May include surrounding context.

python
def _run_cmd(cmd, timeout=120, env=None, **kw):
    """运行子进程,带 timeout。"""
    try:
        return subprocess.run(cmd, timeout=timeout, env=env, **kw)
    except subprocess.TimeoutExpired:
        print(f"       -> 命令超时 ({timeout}s): {' '.join(str(c) for c in cmd[:4])}...")
        return None

Tainted flow: 'dst' from os.environ.get (line 551, credential/environment) → shutil.copy2 (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/parse.py (reported line 552)May include surrounding context.

python
if m:
            pg = int(m.group(1))
            dst = adir / f"page_{pg:02d}.png"
            shutil.copy2(png, dst)
            page_map[pg] = str(dst)

    extracted = {}

Tainted flow: 'json_path' from os.environ.get (line 580, credential/environment) → open (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/parse.py (reported line 581)May include surrounding context.

python
"images": extracted.get(pg, []),
        }
    json_path = adir / "extracted_text.json"
    with open(json_path, "w", encoding="utf-8") as f:
        json.dump(data, f, ensure_ascii=False, indent=2)
    return str(json_path)

Image removal can delete text boxes and their text despite explicit preservation guarantees.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The code removes an entire drawing, pict, or background subtree whenever any descendant matches an image element, potentially deleting text boxes or other text-bearing content in the same container. Text verification normally detects covered text loss, but runs only after the destination has been replaced, leaving a damaged output in standalone use; --no-verify bypasses that detection entirely. This undermines the conversion skill's explicit requirement that the no-image version preserve all text.

Content

No source excerpt is available for this finding.