Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/cutepkq/newton/setupgit clone --depth 1 https://github.com/CUTEPKQ/NEWTONWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00034 | $0.00692 |
| Opus 5 | $0.00017 | $0.00346 |
| Sonnet 5 | $0.00007 | $0.00138 |
| Haiku 4.5 | $0.00003 | $0.00069 |
Grade A, and why
setup scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are helping the user set up the Newton physics-video generation system. Walk through each step below. If a step is already done, skip it and say so.
Step 1 — Conda environment
Check if a conda env named newton exists (conda env list). If not, create it:
conda create -n newton python=3.10 -y
Then activate it. Confirm Python 3.10 is active.
Step 2 — PyTorch
Check if PyTorch is installed (python -c "import torch; print(torch.__version__, torch.cuda.is_available())").
If not installed, detect the CUDA version (nvidia-smi or nvcc --version) and install the matching PyTorch build. Example for CUDA 12.8:
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128
If no GPU is available, install the CPU build.
Step 3 — Python dependencies
pip install -r requirements.txt
Step 4 — API keys
Check if .env exists in the project root. If not, copy from .env.template:
cp .env.template .env
Then ask the user for each API key group, one at a time. Explain what each service is used for, and write the values into .env. The groups are:
- Planner LLM (PLANNER_API_KEY, PLANNER_BASE_URL, PLANNER_MODEL) — the agent brain, any OpenAI-compatible endpoint. Ask the user which provider they use (OpenAI, Azure, or a third-party gateway) and set PLANNER_AUTH accordingly (bearer or api-key).
- Gemini (GEMINI_API_KEY) — used for the video verifier (blind A/B scoring) and optionally for image generation. Get a key at https://ai.google.dev/gemini-api/docs.
- Seedance (SEEDANCE_API_KEY, SEEDANCE_HOST) — the video generation backend. Get access at https://seed.bytedance.com/en/seedance2_0.
- Serper (SERPER_API_KEY) — web image search. Get a key at https://serper.dev/.
- Img_create (IMG_CREATE_API_KEY, IMG_CREATE_BASE_URL, IMG_CREATE_MODEL, IMG_CREATE_API) — image generation/editing. Can share credentials with the Planner or Gemini endpoint. Ask the user which provider they want to use.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 62 lines · 34 tokens per session scan A 55e1df4ac824
setup is a command published in the GitHub repository CUTEPKQ/NEWTON (143 stars, last pushed 1mo ago), licensed MIT. It adds 34 tokens to every session and 692 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
loop
Iteratively fix issues until all resolved or max iterations reached.
OpenSpec: Apply
Implement an approved OpenSpec change and keep tasks in sync.
auto-goal
goal 래퍼 — /goal 생성, 상태 확인, 완료/blocked handoff를 goal tool 또는 slash command로 연결합니다.
dispatcher
Pick the next-best repo to work on across the portfolio — rank free repos, recommend one, claim its lease atomically, and route to the entry command.
implement-task
根据技术方案实施任务并输出实现报告.
propose
Manage parked phase proposals in the audit manifest — list them, materialize one (or all) into live phases, or drop one. Proposals are parked by /audit:init when the user declines (some of) the synthesized plan; materialization is a move, not a re-synthesis.