Back to skill

Security audit

analyze-digital-construction-technologies-rd

Security checks for vulnerabilities and agentic risk

Overview

This skill is an evidence-focused research workflow with a local Word report generator, and I found no hidden persistence, exfiltration, destructive behavior, or privilege escalation.

Installers should understand that this skill helps an agent conduct research and create reports, but the final DOCX is only as reliable as the reviewed evidence package and sources supplied. Use authorized PatSnap or other source tools, verify patent/legal status in official registers for decisions, and avoid using it as engineering certification, legal advice, procurement approval, or investment advice.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • 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 (5)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill is presented as a research and technology-intelligence workflow, but the detected behavior indicates it primarily generates/formats documents and writes schema files rather than performing the promised evidence collection and analysis. This mismatch is dangerous because users may trust outputs as evidence-backed diligence when they are actually generated without the claimed source validation, creating integrity, compliance, and decision-risk issues.

Content

No source excerpt is available for this finding.

Self-Modification

High
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.

Content

Scanner excerpt · scripts/generate_word_report.py (reported line 451)May include surrounding context.

python
def write_schema(path: Path, force: bool) -> None:
    if path.exists() and not force:
        raise ValueError(f"Refusing to replace existing file without --force: {path}")
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_text(json.dumps(SCHEMA_EXAMPLE, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding

The skill references local file generation (scripts/generate_word_report.py) and the static analyzer detected file read/write capabilities, but the manifest does not declare any explicit tool scope or permissions boundary. Undeclared file capabilities increase the risk of unexpected filesystem access, make review harder, and can enable data leakage or modification if the implementation is broader than the documented behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The markdown explicitly instructs the skill to "Produce an English report" for all uses. This imposes a language choice by default and does not offer the user a locale/language option or explain a justified region-specific requirement, which fits the language-policy violation criteria.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 170)May include surrounding context.

md
### Step 6 — Generate deliverables

Prepare a concise executive summary plus methods, evidence tables, limitations and appendices. Generate Word only from a reviewed JSON evidence package using `scripts/generate_word_report.py`. A web/HTML report may be produced only when requested; the source package contains no HTML generator, so do not add one to this package without approval.

### Step 7 — Review