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 instructions/druide67/asiai/agents-mdgit clone --depth 1 https://github.com/druide67/asiaiWrote 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/instructions/druide67/asiai/agents-md)<a href="https://agentmods.dev/instructions/druide67/asiai/agents-md"><img src="https://agentmods.dev/badge/instructions/druide67/asiai/agents-md.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 | $0.01613 | $0.01613 |
| Opus 5 | $0.00807 | $0.00807 |
| Sonnet 5 | $0.00323 | $0.00323 |
| Haiku 4.5 | $0.00161 | $0.00161 |
Grade B, and why
asiai AGENTS.md scanned grade B with 1 finding 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 4d 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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
│ └── power.py # PowerMonitor (sudo powermetrics, GPU/CPU watts) How it starts
The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — asiai
Guidelines for AI agents contributing to this project.
Project
asiai — open-source CLI to manage, benchmark, and monitor local LLM inference on Apple Silicon.
- Language: Python 3.11+, zero external dependencies for the core (stdlib only)
- Target: macOS Apple Silicon only (M1/M2/M3/M4). No Linux, no Windows.
- License: Apache 2.0
Architecture
src/asiai/
├── cli.py # CLI entry point (argparse)
├── doctor.py # Installation and environment diagnostics
├── daemon.py # launchd daemon management (continuous monitoring)
├── engines/ # Engine adapters (5 engines)
│ ├── base.py # ABC InferenceEngine + dataclasses
│ ├── openai_compat.py # Base class for OpenAI-compatible engines (template method)
│ ├── detect.py # Auto-detection (ports 11434, 1234, 8080, 8000)
│ ├── ollama.py # Ollama adapter (native API)
│ ├── lmstudio.py # LM Studio adapter (OpenAI-compatible)
│ ├── mlxlm.py # mlx-lm adapter (OpenAI-compatible, Apple MLX native)
│ ├── llamacpp.py # llama.cpp adapter (OpenAI-compatible, GGUF)
│ └── vllm_mlx.py # vllm-mlx adapter (OpenAI-compatible, MLX)
├── collectors/ # Metric collectors (system, inference, power, macOS native)
│ └── power.py # PowerMonitor (sudo powermetrics, GPU/CPU watts)
├── benchmark/ # Runner + standardized prompts + reporter + regression
│ └── regression.py # Regression detection vs SQLite history
├── storage/ # SQLite (schema, migrations, dataclasses)
├── advisor/ # Hardware-aware recommendations
├── display/ # Renderers (CLI, TUI, Web)
│ ├── cli_renderer.py # CLI rendering (detect, bench, doctor, monitor)
│ ├── formatters.py # Formatting helpers (ANSI, bytes, uptime)
│ ├── tui.py # Textual dashboard (optional)
│ └── tui.tcss # Textual styles
└── web/ # Web dashboard (optional: pip install asiai[web])
├── app.py # FastAPI factory + lifespan + Jinja2
├── state.py # AppState + BenchStatus dataclasses
├── routes/ # Route handlers (dashboard, bench, monitor, history, doctor)
├── templates/ # Jinja2 templates + partials
└── static/ # CSS + JS (ApexCharts config)
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.
- 4d ago First seen · 137 lines · 1,613 tokens per session scan B 564cc25cb18c
asiai AGENTS.md is an instructions file published in the GitHub repository druide67/asiai (11 stars, last pushed 5d ago), licensed Apache-2.0. It adds 1,613 tokens to every session, about $0.0081 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
Rapid-MLX AGENTS.md
AGENTS.md instructions for raullenchai/Rapid-MLX, covering rapid-mlx engineering team, team, working model, required task lifecycle and durable knowledge.
mlx-serve CLAUDE.md
Claude Code instructions for ddalcu/mlx-serve, covering mlx-serve — project context for ai, detail map (read on demand), stack, layout (src/) and building.
apfel CLAUDE.md
Instructions for Arthur-Ficial/apfel, covering apfel - project instructions, the golden goal, core product (this is what apfel is), byproducts (useful, but not the pitch) and readme.md structure rule.
strix-halo-guide AGENTS.md
AGENTS.md instructions for hogeheer499-commits/strix-halo-guide, covering agents.md, core project lens, documentation style, do-not-invent rules and benchmark claim rules.
oh-my-knowledge AGENTS.md
AGENTS.md instructions for lizhiyao/oh-my-knowledge, covering agents.md - agent 入场清单, 开工先做, 自主 cr 与完成定义, code review rules and 测量可比性与公开身份.
delegate-local CLAUDE.md
Instructions for IsmaelMartinez/delegate-local, covering claude.md, what this repo is, commands, architecture and conventions.