quantumlab

Run the /home/bram/work/quantum_lab Python scripts and demos inside the existing venv ~/.venvs/qiskit. Use when asked (e.g., via Telegram/OpenClaw) to run quant_math_lab.py, qcqi_pure_math_playground.py, quantum_app.py subcommands, quantumapp.server, or notebooks under the repo.

This skill legitimately runs a local quantum_lab repo inside a qiskit virtualenv, but it grants the agent the ability to execute arbitrary commands in your repo and venv and to start network services. Before installing: 1) Inspect and trust the contents of ~/work/quantum_lab (or set QUANTUM_LAB_ROOT to a sandbox/copy). 2) Consider using a disposable/sandboxed venv or container rather than your personal ~/.venvs/qiskit. 3) If you do not fully trust autonomous agent actions, disable autonomous invocation or require manual confirmation before running the skill. 4) Be cautious about commands that install packages (pip install -r requirements.txt) or launch servers (python -m quantumapp.server) — these can modify your environment or expose services. If you want to proceed, set explicit QUANTUM_LAB_ROOT and VENV_PATH to controlled paths and review the repo code first.

Type: OpenClaw Skill Name: quantumlab Version: 1.0.0 The skill is classified as suspicious due to the `exec "$@"` command in `scripts/qexec.sh`. While the `SKILL.md` instructions guide the agent to use this script with specific `python` commands, the `exec "$@"` primitive allows for arbitrary command execution if the agent were to be prompted by a malicious user to pass unintended arguments. This represents a broad permission and a risky capability, even though the skill's own instructions do not demonstrate malicious intent. The prompt injection instructions in `SKILL.md` are for defining command aliases and do not aim to subvert the agent for harmful purposes.