Security audit

OPC Product Manager

Security checks for vulnerabilities and agentic risk

Overview

This skill is mostly a coherent product-spec helper, but it gives itself broad instruction priority and can write archived project state and touch contract or invoice context without tight permission scoping.

Install only if you are comfortable with a product-spec skill that can create and update local products/ files and index metadata. Use it in a project-scoped workspace, review any generated build commands before running them, and do not allow contract or invoice lookups unless you explicitly want that business data used for the spec.

A.I.G

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

Findings (1)

T01 · Skill Instruction Hijacking

Warning

Location: SKILL.md:15
Finding: Unscoped Instruction-Priority Override in Skill Prompt
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration

Findings (11)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 98% confidence
Finding: The declared description frames the skill as spec generation, but the body also archives files, updates trackers, runs scripts, and links into contract/invoice domains. This mismatch can mislead users and policy systems into granting or invoking the skill under a narrower trust model than its actual behavior, enabling unexpected data access and file mutation.
Content: No source excerpt is available for this finding.

Direct Prompt Extraction

High

Category: System Prompt Leakage
Confidence: 85% confidence
Finding: Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
Content: Scanner excerpt · SKILL.md (reported line 354)May include surrounding context.
md
--- ## Output Rules - All specs in markdown - Metadata in JSON

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 88% confidence
Finding: The example 'What am I building?' is common everyday language and is presented as a way to invoke or use the skill. Because the README does not clearly limit when this phrase should activate the skill or provide exclusion examples, it risks unintended invocation overlap with ordinary conversation.
Content: No source excerpt is available for this finding.

Undeclared Tool Scope

Medium

Category: MCP Least Privilege
Confidence: 91% confidence
Finding: The skill instructs the agent to read and write files (read_file(...), creating products/{product-slug}/...) but does not declare an explicit tool scope or permission boundary. This creates a least-privilege gap: an orchestrator or reviewer cannot easily tell that filesystem access is required, increasing the risk of unintended file access or writes during execution.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: Several trigger phrases are broad everyday language, which makes unintentional mode switching more likely. Because different modes imply different levels of capability use, vague routing logic can cause the skill to perform file operations or analysis paths the user did not intend.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: Defaulting ambiguous input to Full Spec mode can trigger expansive behavior—reading references, generating artifacts, and potentially writing archived outputs—without clear user intent. In agent systems, ambiguous activation increases the chance of accidental execution and over-collection or mutation of data.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium

Category: Not specified by scanner
Confidence: 88% confidence
Finding: The workflow extends beyond spec writing into persistent archiving, status tracking, and cross-skill data linkage. That scope expansion matters because it turns a content-generation skill into one that can modify project state and aggregate information from multiple domains, increasing attack surface and the chance of unintended side effects.
Content: No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: Reaching into contract and invoice context is not necessary for generating a product spec and violates data minimization. If invoked in environments containing sensitive business records, the skill could unnecessarily expose financial or contractual data to the agent context, broadening confidentiality risk.
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 254)May include surrounding context.
md
6. Output build commands 7. Score complexity No confirmation step — generate and present immediately. ---

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The instruction "Ship in one language" is a natural-language locale policy constraint that directs the skill to avoid multilingual support categorically. Because it does not offer user choice, opt-in, or a region-specific justification, it matches the language/locale policy violation criteria.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 84% confidence
Finding: This code performs a file write to products_dir/INDEX.json, replacing prior contents. Although the module and CLI help mention rebuilding INDEX.json, there is no runtime confirmation, visible log before writing, or explicit warning that an existing index file will be overwritten.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.