Back to skill

Security audit

simplify-code

Security checks for vulnerabilities and agentic risk

Overview

This is a code-refactoring guidance skill with broad but disclosed scope and no hidden execution, persistence, or data exfiltration.

Install this if you want a conservative refactoring workflow. Expect it to read relevant local code/config and run existing project checks when asked to simplify code; review the resulting diff and command report, especially because its activation terms are intentionally broad.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (13)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 115)May include surrounding context.

md
| `references/languages.md` → `languages-tier1.md` / `languages-tier2.md` | Always, for the one language being edited (section only). |

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill advertises on very broad phrases like 'make this cleaner', 'polish recently written code', and multilingual equivalents, which can match many routine developer requests and cause the skill to activate outside a narrowly intended refactoring context. Over-broad routing increases the chance that this skill overrides a more appropriate specialized skill or captures requests where its prescriptive workflow and constraints are not suitable, leading to incorrect handling or unsafe transformations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This JSON manifest is in scope for vague-trigger review, and many prompt/invocation descriptions use generic phrases that overlap with common everyday requests. Because the file does not define activation boundaries, exclusions, or a narrow invocation context, these broad phrasings could cause unintended skill activation in unrelated coding conversations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The natural-language requirement "response in Chinese" and "Response/report is written in Chinese" forces a specific output language as part of skill behavior. This is a language-policy concern because the manifest does not indicate user choice, opt-in handling, or a documented region-specific justification for enforcing Chinese.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

This expectation mandates Chinese output for the skill response, which is a natural-language policy issue when no user-choice mechanism or justified locale constraint is provided. The requirement appears as a hardcoded language rule in the manifest rather than a conditional behavior based on the user's selected language.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

This line imposes a fixed response language in the evaluation criteria. Without explicit opt-in, multilingual fallback, or region-specific rationale, it violates the stated policy against forcing a specific language or locale.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest includes another fixed-language requirement for skill output. Because it does not provide an opt-in path or explain a necessary locale restriction, it is a policy-relevant natural-language constraint.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

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 · evals/evals.json (reported line 168)May include surrounding context.

json
- Functions are verbs (`sendInvoice`), booleans read as predicates (`isActive`, `hasItems`), classes are nouns.
- Name length proportional to scope size.
- **Clarity at the call site**: prefer APIs that read well where they are *used* (`users.filter(active)` beats cryptic flags).
- **Don't** rename public API symbols without asking.

## 2. Functions
- Small enough to do **one thing**; if you can extract a function with a meaningful name *and* the result is deeper (see §3), extract — don't create shallow one-liners that only add hops.

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 · references/rules.md (reported line 18)May include surrounding context.

md
- Functions are verbs (`sendInvoice`), booleans read as predicates (`isActive`, `hasItems`), classes are nouns.
- Name length proportional to scope size.
- **Clarity at the call site**: prefer APIs that read well where they are *used* (`users.filter(active)` beats cryptic flags).
- **Don't** rename public API symbols without asking.

## 2. Functions
- Small enough to do **one thing**; if you can extract a function with a meaningful name *and* the result is deeper (see §3), extract — don't create shallow one-liners that only add hops.

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 · references/rules.md (reported line 97)May include surrounding context.

md
- Keep try blocks small; extract the body of a try into its own function.
- Handle every promise rejection / future error.
- **Hidden limits are behavior:** when replacing a fixed buffer/array with a growable container (`std::string`, `Vec`, list), the old size limit and its error are observable. Keep them or ask.
- These are targets for new code; in a refactor, **golden rules 1–2 win**: don't convert error codes/strings/panics/null returns to exceptions/Result on a public API without approval.
- **RAII / `with` / `using` / `defer` / try-with-resources**: acquire and release resources via language constructs so cleanup runs on every path (success, error, early return).

## 11. Objects, classes & SOLID

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The file includes parallel Chinese-language content and headings, but does not state that language selection is optional or explain why Chinese is required. Under the policy rule, forcing a specific language or locale without user opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The instruction says to write the final output "in the user's language," which imposes a language/locale behavior without offering a user choice or opt-in. Under the policy, forced language behavior is a natural-language policy concern unless the skill offers a choice or clearly documents a justified locale constraint.

Content

No source excerpt is available for this finding.