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/ktwu01/benchmark-radar/agents-mdgit clone --depth 1 https://github.com/ktwu01/benchmark-radarWrote 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/ktwu01/benchmark-radar/agents-md)<a href="https://agentmods.dev/instructions/ktwu01/benchmark-radar/agents-md"><img src="https://agentmods.dev/badge/instructions/ktwu01/benchmark-radar/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 | $0.03385 | $0.03385 |
| Opus 5 | $0.01692 | $0.01692 |
| Sonnet 5 | $0.00677 | $0.00677 |
| Haiku 4.5 | $0.00338 | $0.00338 |
Grade A, and why
benchmark-radar AGENTS.md 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 yesterday.
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 — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Instructions
Glob rule: showcase and UI communication
Applies to README*, docs/**, .github/ISSUE_TEMPLATE/**, site/**, and
any report, launch note, TLDR, screenshot, GIF, demo, dashboard, or UI surface.
- Treat what is shown as part of the work. What was done and what is displayed are both important; in many communication surfaces, what is displayed is more important because it is the receiver's entry point.
- Start from the receiver's perspective, not the implementer's. Ask what the reader most wants to know, what will help them decide quickly, and what is most worth remembering or sharing.
- Do not let engineering effort bury the message. Data work and implementation details often take most of the time, but reports and TLDRs should foreground the result, implication, and decision-useful signal before the process.
- Prefer strong information hierarchy, plain language, concrete examples, screenshots, short GIFs, and compact summaries that make the work easy to scan, review, forward, or explain upward.
Example: simplify badge copy and keep its style
Before:
<p align="center">
<a href="https://koutian.is-a.dev/benchmark-radar/"><img alt="Benchmark records collected" src="https://img.shields.io/endpoint?url=https%3A%2F%2Fkoutian.is-a.dev%2Fbenchmark-radar%2Fdata%2Frecords-badge.json&style=for-the-badge"></a>
<a href="https://koutian.is-a.dev/benchmark-radar/data/radar.json"><img alt="Download dataset" src="https://img.shields.io/badge/Dataset-download%20JSON-2f81f7?style=for-the-badge&logo=json&logoColor=white"></a>
<a href="https://x.com/ktwu01"><img alt="X" src="https://img.shields.io/badge/X-%40ktwu01-000000?style=for-the-badge&logo=x&logoColor=white"></a>
<a href="https://www.linkedin.com/in/ktwu01"><img alt="LinkedIn" src="https://img.shields.io/badge/LinkedIn-Koutian%20Wu-0A66C2?style=for-the-badge&logo=linkedin&logoColor=white"></a>
<a href="https://scholar.google.com/citations?user=s9w1k-cAAAAJ&hl=en"><img alt="Google Scholar" src="https://img.shields.io/badge/Google%20Scholar-Koutian%20Wu-4285F4?style=for-the-badge&logo=googlescholar&logoColor=white"></a>
</p>
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.
- yesterday Changed · +26 lines · +418 tokens per session ef3cc236ac97
- 5d ago First seen · 213 lines · 2,967 tokens per session scan A 94485ad1a3e0
benchmark-radar AGENTS.md is an instructions file published in the GitHub repository ktwu01/benchmark-radar (145 stars, last pushed yesterday), licensed MIT. It adds 3,385 tokens to every session, about $0.0169 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 instructions, from other repositories
langwatch CLAUDE.md
Claude Code instructions for langwatch/langwatch, covering langwatch, before you implement anything, development environment, running with no container runtime and local dev by hostname — thuishaven / portless (recommended).
AutoRAG AGENTS.md
AGENTS.md instructions for Marker-Inc-Korea/AutoRAG, covering autorag — pi-powered librarian agent, git workflow (binding), before any pr review, pr work, or new feature, after the work is finished and developer commands.
Hegelion CLAUDE.md
Instructions for Hmbown/Hegelion, covering claude.md, what is hegelion, build & development commands, install dependencies (uv recommended) and run all tests.
Hegelion AGENTS.md
Instructions for Hmbown/Hegelion, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
AutoRAG copilot-instructions.md
Copilot instructions for Marker-Inc-Korea/AutoRAG, covering autorag - ai-powered rag pipeline optimization, working effectively, prerequisites and system setup, platform dependencies and install java 17 (required).
helicone AGENTS.md
AGENTS.md instructions for Helicone/helicone, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.