The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to download and immediately execute scripts from https://nextmovetheory.com/install.sh and https://nextmovetheory.com/install.ps1 using curl | bash and Invoke-RestMethod | Invoke-Expression patterns. This bypasses security reviews by allowing unverified remote content to execute arbitrary shell commands.
[EXTERNAL_DOWNLOADS]: The skill relies on resources from nextmovetheory.com, which is not a recognized trusted vendor or well-known service, increasing the risk of supply-chain attacks.
[DATA_EXFILTRATION]: The skill reads the contents of the local file .nmt-version and appends it to a network request sent to https://nextmovetheory.com/version. This pattern demonstrates the exfiltration of local file data to an external server.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes unverified output from a remote installer.
Ingestion points: Installer script output and git status results in the SKILL.md file.
Boundary markers: None present; the agent is instructed to report the output directly.
Capability inventory: Shell execution capabilities (bash) are present in the skill.
Sanitization: No sanitization or validation of the remote output is performed before it is processed by the agent.

nmt-upgrade