Back to skill

Security audit

Chrome Extension Maintainer

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed review workflow for maintaining Chrome extensions and does not include executable code, hidden persistence, or automatic publishing behavior.

Installers should treat this as a release-review rubric: it may ask the agent to inspect extension code, metadata, privacy claims, analytics exports, and local release artifacts, and to run repo-provided checks. Do not approve actual uploads, store submissions, unpublishing, rollbacks, registry publishing, or broad promotion unless you intentionally want that action.

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
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 92)May include surrounding context.

md
- `SKILL.md` is valid and has concrete workflow, rules, and validation guidance.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 137)May include surrounding context.

md
- `SKILL.md` is valid and has concrete workflow, rules, and validation guidance.

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · references/chrome-extension-maintenance-playbook.md (reported line 131)May include surrounding context.

md
- Narrow one-click workflows over broad dashboards.
- `activeTab` and click-triggered scripting over persistent host permissions.
- Local-first output with optional remote improvement only after explicit user action.
- Clear state labels: ready, reading, improving, copied, exported, failed.
- Small versioned releases with one measurable behavior change.
- Locale expansion based on actual language/region usage and support burden.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

L012 states the skill does not bundle an uploader or publisher and should rely on repo-local tools, implying a review-oriented workflow rather than publication mechanics. Later guidance at L081-L096 and L137 includes publication-oriented steps such as registry auth checks and pre-publish validation for skill packages, which softens and partially contradicts the earlier 'does not bundle' framing.

Content

No source excerpt is available for this finding.