gitops-security

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, potentially untrusted data from Git repositories and Kubernetes clusters, creating an attack surface where malicious data could influence agent behavior.
  • Ingestion points: The skill describes cloning external Git repositories (git clone, flux create source) and querying Kubernetes APIs (kubectl get) in SKILL.md and guides/gitops-security-playbook.md.
  • Boundary markers: The instructions do not explicitly mandate the use of boundary markers or safety delimiters when the agent processes content retrieved from these external sources.
  • Capability inventory: The skill utilizes tools like kubectl, helm, argocd, flux, git, and curl, which provide extensive capabilities for file system manipulation, network operations, and command execution across multiple files.
  • Sanitization: There is no documentation of automated sanitization or validation of data retrieved from external sources before it is processed by the agent.
  • [EXTERNAL_DOWNLOADS]: The skill downloads and applies Kubernetes manifests from remote sources during setup.
  • Evidence: guides/gitops-security-playbook.md includes commands to fetch manifests from raw.githubusercontent.com/argoproj/argo-cd/v2.4.0/manifests/install.yaml and raw.githubusercontent.com/kyverno/kyverno/main/config/install.yaml. These sources are well-known and official repositories for the respective open-source projects.
  • [COMMAND_EXECUTION]: The skill relies on the execution of numerous shell commands via specialized CLI tools to perform its intended functions.
  • Evidence: Tools such as kubectl, helm, argocd, flux, tkn, git, sops, kubeseal, and vault are used throughout the playbooks and payload documentation. This behavior is consistent with the skill's primary purpose as a security testing and red-teaming resource.
  • [DYNAMIC_EXECUTION]: The skill includes code snippets for runtime execution, such as Python scripts and shell-injected jobs.
  • Evidence: payloads.md contains a Python snippet for forging JWT tokens and various Kubernetes Job manifests that execute shell scripts (e.g., in §3.2 and §10.1). These are presented as attack PoCs and educational examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 04:36 PM