Borrowing it
Nothing to install: this file belongs to saski/arnesto. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/saski/arnesto/main/.agents/skills/codex-model-routing/SKILL.mdgit clone --depth 1 https://github.com/saski/arnestoWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/saski/arnesto/codex-model-routing)<a href="https://agentmods.dev/skills/saski/arnesto/codex-model-routing"><img src="https://agentmods.dev/badge/skills/saski/arnesto/codex-model-routing.svg" alt="Measured on agentmods" height="20"></a>What 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 5.1 | $0.00098 | $0.01302 |
| Opus 5 | $0.00049 | $0.00651 |
| Sonnet 5 | $0.00020 | $0.00260 |
| Haiku 4.5 | $0.00010 | $0.00130 |
Grade A, and why
codex-model-routing 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 today.
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 — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codex Model Routing
Recommend the least expensive route likely to produce an accepted result. This is an on-demand advisory policy for Codex/OpenAI only. Preserve existing model-routing criteria for Claude, Gemini, Hermes, OmniRoute, OpenCode, and other providers. Do not change global configuration, model defaults, symlinks, hooks, or MCPs as part of a recommendation. A recommendation is not evidence that the active model, effort, or Fast setting changed.
Choose a route
- Identify the deliverable, acceptance check, ambiguity, risk, dependencies, and value of reduced latency. Use supplied context; ask only when a missing constraint materially changes the choice.
- Select a model from the policy below. Terra Medium is the default recommendation, not a configuration change or an obligation to use Terra for a clearly harder task.
- Select effort separately from Fast. Explain the cheapest plausible route and the evidence that would justify escalation.
- Return the model, effort, Fast on/off, brief reason, acceptance check, and stop/escalation condition. For a spending review, also report measurement gaps and a small comparison plan.
| Work | Starting model and effort |
|---|---|
| Clear, repeatable extraction, classification, transformation, or mechanical checks with defined input/output | Luna Light/Low; Medium if it needs several bounded steps |
| Everyday bounded implementation, documents, reporting, or analysis with tools | Terra Medium |
| Several sources, dependent steps, edge cases, or explicit trade-offs within a clear specification | Terra High |
| Open-ended diagnosis, architecture, cross-repository complexity, consequential delivery, or deep risk review | Sol Medium or High; High for difficult diagnosis or risk |
| Exceptionally difficult end-to-end workflows combining code, research, computer use, and sustained judgment | Astra Light/Low or Medium; justify why Sol is insufficient |
Do not infer that a more expensive model is better from spend per turn alone. For the source rationale and measurement limits, read policy evidence.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- today First seen · 106 lines · 98 tokens per session scan A 9f5f90a03d86
codex-model-routing is a skill published in the GitHub repository saski/arnesto (5 stars, last pushed today), licensed Unlicense. It adds 98 tokens to every session and 1,302 once invoked, about $0.0005 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 2026-09-06.
Other skills, from other repositories
fs-creative-voltage
OpenDesign's seed pitch: the open, local alternative to closed AI design — why now, the wedge, and the ask. Built as a decision-grade fundraising pitch deck for pre-seed & seed VCs.
fs-notebook-tabs
A computer-science capstone: an on-device ML keyboard that predicts next words privately — problem, method, evaluation, and defense answers. Built as a decision-grade coursework defense deck for professor, defense committee.
deepstream-sop
Use this skill when building, deploying, evaluating, debugging, or measuring latency for the DeepStream SOP Inference Microservice — a GPU-accelerated FastAPI service that detects whether operators perform assembly-line steps in order via event boundary detection (GEBD) plus VLM classification. Trigger even if the…
digital-health-clinical-asr-finetune
Stage 4 of the Clinical ASR Flywheel. Use when priority KER is above 0.3 to run stock NeMo SFT on Parakeet TDT v2 and offline cycle N+1 re-eval. NOT for generic word boosting (use /finetune-asr).
nemo-automodel-distributed-training
Guide for selecting and configuring distributed training strategies in NeMo AutoModel, including FSDP2, Megatron FSDP, DDP, and parallelism settings.
aiq-research
Use when asked to run deep research or AI-Q research through a reachable NVIDIA AI-Q Blueprint backend.