Back to skill

Security audit

Publish Website Traefik

Security checks for vulnerabilities and agentic risk

Overview

This skill is meant to publish and delete Docker-hosted static sites, but its Docker and deletion scripts are under-scoped and could be unsafe if made runnable.

Review this skill carefully before installing. It is not deceptive, but it expects Docker authority and includes deletion behavior; only use it in an isolated environment after adding strict subdomain validation, a confined deployment directory, protected state-file permissions, and an explicit delete confirmation or backup workflow.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/deploy_site.sh:1
Finding

Unvalidated Subdomain Allows Docker Compose YAML Injection

Content
View full analysis
\"$TEMP_DIR/docker-compose.yml\" version: '3.8' services: nginx: image: nginx:alpine container_name: \"${SUBDOMAIN}-web\" labels: - \"traefik.enable=true\" - \"traefik.http.routers.${SUBDOMAIN}-router.rule=Host(\`${FULL_DOMAIN}\`)\" - \"traefik.http.routers.${SUBDOMAIN}-router.entrypoints=websecure\" - \"traefik.http.routers.${SUBDOMAIN}-router.tls=true\" - \"traefik.http.routers.${SUBDOMAIN}-router.tls.certresolver=le\" - \"traefik.http.services.${SUBDOMAIN}-service.loadbalancer.server.port=80\" networks: - web networks: web: external: true EOF pushd \"$TEMP_DIR\" || { echo \"Failed to change directory to $TEMP_DIR\"; exit 1; } docker compose up -d || { echo \"Failed to start docker compose services\"; popd; exit 1; } ``` ### Technical Analysis The second positional argument is accepted as `SUBDOMAIN` without validation and interpolated directly into a generated Docker Compose YAML document. Shell quoting around the variable does not protect the structure of the generated YAML because variable expansion occurs while processing the heredoc. A subdomain containing quote characters, line breaks, or YAML syntax can potentially terminate an existing scalar and inject additional Compose properties. Depending on the crafted structure, an attacker could attempt to introduce properties such as: - Host filesystem bind mounts - An attacker-controlled container command or entry point - Additional environment variables - Privileged container mode - Ho ...[truncated 1792 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/delete_site.sh:1
Finding

Untrusted Deployment State Controls Recursive Directory Deletion

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The primary behavior shown matches the declared deployment purpose: it creates a Docker Compose configuration for nginx, configures Traefik labels for the requested subdomain, starts the container, copies static files into nginx, and produces a URL under sites.friendify.cloud. However, the declared description says the skill provides deploy, list, and delete functions, while this code chunk only implements deployment. Since the task is to compare the supplied code chunk itself, that alone is not necessarily a mismatch unless the code contradicts the description. The meaningful undeclared behavior is that it persists deployment metadata to a hard-coded workspace file path, which is access to an additional local resource not described. Aside from that, the code is aligned with the stated purpose.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file describes a delete command that stops and removes a Docker container and its resources, which is a destructive operation affecting service availability. While the action is named 'delete,' the documentation does not include a clear warning or caution about irreversibility, downtime, or verifying the target subdomain before proceeding.

Content

No source excerpt is available for this finding.