Exec Local

Execute system-level shell commands directly within the thundarr-gpu container for monitoring and remote orchestration tasks.

Security Scan

The skill legitimately provides arbitrary shell execution inside the container (which matches its name), but that capability is extremely powerful and the instructions give no meaningful limits — this is coherent with the stated purpose but disproportionate in privilege and risk for most users.

Purpose & Capability

The name/description match the provided tooling: an 'exec' tool that runs arbitrary shell commands inside the Thundarr container. Mentioning remote orchestration via an SSH script is plausible, but the skill does not declare or request SSH keys/configs even though those would commonly be required for the described remote tasks.

Instruction Scope

SKILL.md explicitly enables arbitrary command execution and gives no runtime restrictions. Although consistent with the stated purpose, this scope permits reading files, environment variables, network access, and other actions not limited by the skill — enabling easy exfiltration or lateral movement if misused.

Install Mechanism

Instruction-only skill with no install spec or downloaded code. This minimizes installation risk (nothing written by the skill itself), but runtime risk remains because commands executed by the agent will run in the container.

Credentials

The skill declares no environment variables or credentials, which is consistent with an in-container exec tool. However, arbitrary shell access effectively gives the agent access to any env vars, config files, mounted secrets, SSH keys, or sockets present in the container — a disproportionate level of access relative to what most users would expect from a small utility.

Persistence & Privilege

The skill is user-invocable and allows autonomous model invocation (disable-model-invocation is false). Autonomous invocation combined with unrestricted shell execution increases the blast radius: the agent could run commands without explicit user approval. 'always' is not set, which limits forced inclusion, but the autonomous capability is still concerning given the privileges.

What to consider before installing

This skill gives the agent unrestricted ability to run shell commands inside the container — it can read files, environment variables, network endpoints, and any mounted host artifacts (SSH keys, docker socket, config files). Only install if you fully trust the skill and the agent's autonomy. Consider: 1) Disable autonomous invocation (set disable-model-invocation = true) so the agent cannot run commands without your approval; 2) Remove or isolate sensitive mounts/credentials from the container (SSH keys, /var/run/docker.sock, host config directories); 3) Limit network egress from the container or firewall outbound connections; 4) Prefer manual invocation: require explicit commands from a human rather than letting the model decide; 5) Audit container file system and logs after use; 6) If you need limited monitoring/orchestration, prefer more narrowly scoped tools that expose specific APIs rather than an open shell. If you are unsure, treat this skill as high-risk and avoid installing it.

Thundarr Execution Bridge (Local)

The exec_local skill provides a direct interface for executing system-level shell commands within the thundarr-gpu container environment on Garuda.

Core Functionality

• Direct Shell Access: Allows for low-latency execution of binary tools and scripts.
• Remote Orchestration: Designed to be used as a jump-point for managing the Fedora UM250 host via the thundarr-remote SSH script.
• Resource Monitoring: Can be used to check local container health, process tables, and network availability between Docker nodes.

Security Note

This tool executes commands with the permissions of the container user. Ensure all inputs are sanitized when used in automated workflows.