Security audit

Jenkins

Security checks for vulnerabilities and agentic risk

Overview

This Jenkins skill does what it says: stores project/job configuration locally and uses Jenkins credentials to view and trigger CI builds with a confirmation step.

Install only if you intend to let the agent use your Jenkins API token to inspect jobs and trigger builds. Review PROJECTS.md because it will store Jenkins URLs and repository URLs, and confirm the selected project carefully before allowing a build trigger.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (6)

Description-Behavior Mismatch

Medium

Confidence: 86% confidence
Finding: The skill advertises Jenkins REST interactions but also includes local configuration management of PROJECTS.md, expanding its effective scope beyond the stated purpose. This can cause unintended local file modification and persistence of infrastructure metadata without clear user expectation, which is a security-relevant capability expansion even if not overtly malicious.

Context-Inappropriate Capability

Low

Confidence: 80% confidence
Finding: Collecting and storing source repository URLs is not necessary for the core action of triggering Jenkins jobs and therefore introduces unnecessary data collection. Even low-sensitivity metadata can aid reconnaissance by mapping internal development assets and increasing the skill's access to unrelated project information.

Context-Inappropriate Capability

Low

Confidence: 83% confidence
Finding: Automatically inspecting workspace context and grepping local files to infer a project goes beyond a narrowly scoped Jenkins action and may access local context the user did not intend to expose. This increases the risk of unintended data access and accidental project selection, especially when invocation phrases are broad.

Vague Triggers

High

Confidence: 92% confidence
Finding: The trigger phrases are overly broad, including generic terms like 'build' and 'jenkins', which can cause accidental invocation of a skill capable of triggering remote CI/CD jobs. In this context, unintended activation is dangerous because the skill can perform side-effecting operations against Jenkins after only limited interaction.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill does not clearly warn users that configuration actions will write to PROJECTS.md, which can lead to silent persistence of project and infrastructure metadata. Hidden local state changes are security-relevant because they alter workspace files and may affect later automation behavior.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill lacks a clear warning that triggering a Jenkins build initiates remote CI/CD jobs that may deploy code, consume infrastructure, access secrets, or modify environments. In a CI/CD context, build triggers are inherently side-effecting and can have significant operational and security consequences if users are not properly informed.

Static analysis

No suspicious patterns detected.