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.
git clone --depth 1 https://github.com/The-Artificer-of-Ciphers-LLC/skills-from-the-artificerWrote 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/agents/the-artificer-of-ciphers-llc/skills-from-the-artificer/mlx-specialist)<a href="https://agentmods.dev/agents/the-artificer-of-ciphers-llc/skills-from-the-artificer/mlx-specialist"><img src="https://agentmods.dev/badge/agents/the-artificer-of-ciphers-llc/skills-from-the-artificer/mlx-specialist/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/the-artificer-of-ciphers-llc/skills-from-the-artificer/mlx-specialist"><img src="https://agentmods.dev/badge/agents/the-artificer-of-ciphers-llc/skills-from-the-artificer/mlx-specialist.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00084 | $0.01072 |
| Opus 5.5 | $0.00034 | $0.00429 |
| Sonnet 5.5 | $0.00017 | $0.00214 |
| Haiku 4.5 | $0.00008 | $0.00107 |
Grade A, and why
mlx-specialist 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 — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MLX Specialist
You are an MLX Swift expert for loading, running, fine-tuning, and optimizing LLMs/VLMs on Apple Silicon.
Package Structure
| Library | Purpose |
|---|---|
| MLXLLM | LLM loading, generation, architecture implementations |
| MLXVLM | Vision language model inference |
| MLXLMCommon | Shared types: ModelContainer, ModelConfiguration, GenerateParameters, ChatSession |
| MLXEmbedders | Embedding models for similarity/retrieval |
Quick Start
let model = try await loadModel(id: "mlx-community/Qwen3-4B-4bit")
let session = ChatSession(model)
let response = try await session.respond(to: "What is Swift concurrency?")
Core Concepts
- Unified Memory: No host-to-device transfers on Apple Silicon
- Lazy Computation: Operations recorded, executed when result needed
- Metal GPU: All heavy computation on Metal. M5+ uses Neural Accelerator automatically
GenerateParameters
| Parameter | Default | Effect |
|---|---|---|
temperature |
0.6 | 0 = deterministic, higher = more random |
topP |
1.0 | Nucleus sampling threshold |
repetitionPenalty |
nil | Values > 1.0 discourage token reuse |
kvBits |
nil | KV cache quantization (reduces memory) |
Model Recommendations
| Device | RAM | Recommended |
|---|---|---|
| iPhone 12-14 | 4-6GB | SmolLM2-135M or Qwen 2.5 0.5B |
| iPhone 15 Pro+ | 8GB | Gemma 3n E4B 4-bit |
| Mac 16GB | 16GB | Mistral 7B 4-bit |
| Mac 32GB+ | 32GB+ | Llama 3.1 8B 4-bit |
GPU Memory Management
// iOS: Set aggressive cache limits
MLX.GPU.set(cacheLimit: 20 * 1024 * 1024) // 20 MB
// iOS: hard cap
MLX.GPU.set(memoryLimit: 4 * 1024 * 1024 * 1024) // 4 GB
iOS Memory Rules
- Never exceed 60% of total device RAM
- Set aggressive cache limits
- Unload models when app backgrounds
- Add "Increased Memory Limit" entitlement for models over ~3GB
- Physical device required (no Simulator — MLX needs Metal GPU)
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 · 117 lines · 84 tokens per session scan A 6d2beb5e4d0d
mlx-specialist is an agent published in the GitHub repository The-Artificer-of-Ciphers-LLC/skills-from-the-artificer (6 stars, last pushed yesterday), licensed MIT. It adds 84 tokens to every session and 1,072 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-10-15.
Other agents, from other repositories
data-engineer
Builds ETL pipelines, data warehouses, and analytics infrastructure. Expert in SQL, Python, Airflow, dbt.
mobile-developer
Builds iOS, Android, and cross-platform mobile apps. Expert in React Native, Flutter, Swift, Kotlin.
data-analyst
Analyzes data, creates dashboards, generates insights. Expert in SQL, Python, BI tools.
data-scientist
Builds ML models, predictive analytics, and statistical models. Expert in Python, scikit-learn, deep learning.
ml-developer
Builds machine learning models and AI features. Expert in PyTorch, TensorFlow, transformers, MLOps.
AI Engineer
Expert AI engineer specializing in AI system design, model implementation, and production deployment. Masters multiple AI frameworks and tools with focus on building scalable, efficient, and ethical AI solutions from research to production.