Back to skill

Security audit

Pkg Paper Rewriter Claw

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because scanners found that its rewriting and integrity checks may alter protected terms or numbers, and the review was partial.

The verdict is suspicious, so consider these reported content-integrity weaknesses and the partial scan coverage before installing. In particular, changed terminology or numeric values could require careful checking of any output.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (12)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The concern is not terminology unification itself, which is explicitly documented, but replacement within protected identifiers and other non-term content. Unrestricted substring replacement can corrupt references or scientific data despite the preservation guarantee; opt-in execution and review instructions reduce exposure but do not enforce that guarantee.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

The examples explicitly show check_terms.py --unify replacing 心力衰竭 with 心衰 and 病人 with 患者 in an output document. Although optional and presented as requiring review, this changes terminology rather than preserving it untouched. The document conversion and cleanup workflows otherwise fit the declared purpose. The chunk contains usage examples rather than implementation, so the assessment concerns the behavior those examples explicitly advertise.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 146)May include surrounding context.

md
Worked end-to-end examples: `references/examples.md`. Unified exit codes,

The --unify operation rewrites terminology and can alter protected identifiers despite the manifest's preservation guarantee.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

unify_text applies literal substring replacements across the entire manuscript without excluding identifiers, references, numbers, or larger words. For example, the built-in ID variant can match inside PMID and be replaced when identifier is dominant, corrupting a protected reference label; verification is only recommended after output is written.

Content

No source excerpt is available for this finding.

The bundle's purported complete write-audit manifest records only bundle artifacts, omitting other files written during the same run.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest explicitly claims to list all writes during the run, but its registration mechanism covers only bundle artifacts. Other enabled outputs, including the cleaned manuscript, automatic terminology draft, revision logs, suggestions, handoff, and HTML report, are omitted, potentially misleading users or agents relying on it as permission or audit evidence. This is an audit-completeness weakness, not evidence of unauthorized writes or exfiltration.

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/reporter.py (reported line 460)May include surrounding context.

python
smps = "".join("<li>%s</li>" % esc(s) for s in c.get("samples", [])[:6])
        rows.append("<details><summary>%s ×%d%s</summary>%s<ul>%s</ul></details>"
                    % (esc(c.get("label", cid)), c.get("count", 0),
                       " [可自动修]" if cid in __import__("hxt_core").AUTO_FIXABLE else "",
                       rat, smps))
    hit_html = ("<div class='card'><h2>命中明细(改稿口径,L=原文行号)</h2>"
                + ("".join(rows) if rows else "<p>未命中任何特征。</p>") + "</div>")

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/reporter.py (reported line 486)May include surrounding context.

python
for cid, c in sorted(cats.items(), key=lambda kv: -kv[1].get("count", 0)):
        label = c.get("label", cid)
        cnt = c.get("count", 0)
        if cid in __import__("hxt_core").AUTO_FIXABLE:
            act_auto.append("<li>%s ×%d → <code>python scripts/transform.py 原稿 -o 改稿.txt</code>(机械层自动处理)</li>"
                            % (esc(label), cnt))
        elif cid.startswith(("chatbot", "cutoff", "model_artifact")):

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/selfcheck.py (reported line 65)May include surrounding context.

python
bad = []
    for m in mods:
        try:
            __import__(m)
        except Exception as e:  # noqa: BLE001 — 自检要兜住任意导入失败
            bad.append("%s(%s)" % (m, str(e)[:40]))
    check("S2.模块导入", not bad, "、".join(bad))

The documented stdout mode never emits the cleaned manuscript.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The usage documentation at L014–L015 and argument help at L208 promise cleaned text on stdout when no output file is supplied. However, L263–L268 write the transformed text only when --output is provided, while L284–L301 print a JSON diagnostic report instead; with --quiet, no cleaned text is emitted at all. Consequently, the documented stdin/stdout workflow cannot deliver the rewritten manuscript.

Content

No source excerpt is available for this finding.

The numeric integrity guard accepts changes to scientific values despite the manifest's promise that numbers survive untouched.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The numeric key uses round(float(tok), 6), causing distinct scientific values such as 0.0000001 and 0.0000002 to compare equal; floating-point conversion can also collapse distinct large integers. Consequently, modified data can evade the numeric-integrity violation and receive a passing result, since contextual or sentence-change warnings do not block acceptance by default. This is particularly consequential in an academic rewriting tool that promises to preserve numerical data.

Content

No source excerpt is available for this finding.

The CLI error policy mandates Chinese-language diagnostics without user locale choice.

Low
Category
Output Handling
Confidence
92% confidence
Finding

The centralized exception policy specifies that every user-facing exception produces two lines in Chinese. This fixes the diagnostic language without offering user opt-in or documenting a justification for the locale restriction.

Content

No source excerpt is available for this finding.

CLI error-handling policy requires Chinese output without offering a language choice.

Low
Category
Output Handling
Confidence
96% confidence
Finding

The docstring explicitly requires all CLIs to explain errors and remedies in Chinese; the centralized CLI entry point reiterates this requirement at L2007. No language opt-in or locale selection is provided, and targeting nontechnical users does not justify imposing a particular language.

Content

No source excerpt is available for this finding.