T03 · Remote Payload Retrieval and Execution
- Location
SKILL.md:80- Finding
Unverified Remote Installer Executed Through a Shell Pipeline
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This deployment skill is purpose-aligned, but it can install mutable external deployment tools and run a remote shell installer before deploying real projects.
Review this skill before installing. It is not clearly malicious, but use it only if you are comfortable with an agent installing deployment CLIs globally and running a Fly.io remote installer. Prefer manually installing trusted, pinned CLIs first, and require explicit confirmation for any deployment target and environment.
SKILL.md:80Unverified Remote Installer Executed Through a Shell Pipeline
SKILL.md:56Unpinned Global Installation of Deployment CLI Packages
Piping a remote script directly into the shell executes unverified code from the network with the current user's privileges. If the remote server, transport, or script is compromised, this can result in arbitrary code execution, credential theft, persistence, or modification of the local environment used for deployment.
# Check if fly CLI is installed
which fly || curl -L https://fly.io/install.sh | sh
# If no fly.toml, launch new app
fly launch --yes --no-deploy
The | sh construct is a dangerous command-chaining pattern because it turns downloaded content into immediate shell input without validation. In this deployment skill, that risk is amplified because the command may run automatically during tool setup, enabling pre-deployment compromise of the host and any deployment credentials accessible to the CLI.
# Check if fly CLI is installed
which fly || curl -L https://fly.io/install.sh | sh
# If no fly.toml, launch new app
fly launch --yes --no-deploy
The trigger list includes broad phrases such as "ship it" and "go live," which can cause the deploy skill to activate in contexts where the user did not intend an actual deployment action. In a skill that can perform real production-affecting operations, ambiguous activation increases the chance of unintended deployments or follow-on execution of risky commands.
The core behavior says to act when the user says generic words like "deploy" or "ship," which is overly permissive for a capability that can change external infrastructure. Because this skill performs operational actions, ambiguous activation is more dangerous than in a read-only skill and can lead to accidental deployments or unauthorized state changes if invoked by loosely matching user text.