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/devforth/adminforth/agents-mdgit clone --depth 1 https://github.com/devforth/adminforthWhat 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 | $0.01395 | $0.01395 |
| Opus 5 | $0.00698 | $0.00698 |
| Sonnet 5 | $0.00279 | $0.00279 |
| Haiku 4.5 | $0.00139 | $0.00139 |
Grade A, and why
adminforth 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 — 230 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Package manager
All packages and projects in this repo use pnpm and not npm.
Howeverer internally (e.g. in codeInjector) adminforth still supports both npm and pnpm style install commands, so users of framework itself can use it with either package manager. But in all dev demo/live demo, plugins, adapters, and documentation, we use pnpm as the standard.
Package names rules
All adapters and plugins always have @adminforth/ prefix in their package name, followed by short lowercase kebab-case plugin/adpater slug.
Every plugin has at least one Docusaurus docs page, which should use the path /docs/tutorial/Plugins/<plugin-slug>/.
Same for adapters, but with /docs/tutorial/Adapters/<adapter-slug>/ path.
Page names in docusarus should be human readabale. We should not use AuditLog but instead we should have Audit Log via whitespace.
General engineering rules
Write code as if the system contracts are already defined and trusted.
Do not add defensive checks “just in case” when the contract, type, schema, backend response, or controlled internal API already guarantees the shape of the data.
Prefer simplicity, clarity, and consistency over paranoia.
Follow these principles strictly:
- DRY
- SOLID
- YAGNI
- Keep code minimal
- Trust typed contracts
- Avoid duplicate validation
- Avoid speculative fallback logic
- Avoid inline regex literals when reused or non-trivial
Trust the contract
If backend and frontend are part of the same system, and the backend explicitly guarantees a response shape, do not re-validate every field on the frontend.
Bad:
- backend returns a strict object
- frontend then checks every field for
undefined, wrong type, empty string, wrong enum, etc. - frontend adds fallback branches for impossible states
Good:
- backend owns validation
- shared types or schemas define the contract
- frontend consumes the contract directly
- frontend only handles real UI states, not imaginary protocol corruption
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 First seen · 230 lines · 1,395 tokens per session scan A 6bfb3b96d744
adminforth AGENTS.md is an instructions file published in the GitHub repository devforth/adminforth (390 stars, last pushed 4d ago), licensed MIT. It adds 1,395 tokens to every session, about $0.0070 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
Adminator-admin-dashboard CLAUDE.md
Instructions for puikinsh/Adminator-admin-dashboard, covering claude.md, architecture (2026 redesign), file layout, page anatomy and widget mount points.
Cyber-AI-Forge AGENTS.md
Instructions for tanghaojie/Cyber-AI-Forge, covering ai 协作规则, 人类修改优先与冲突处理, git 安全门禁与 ai 提交标记, 开始修改前的暂存区硬门禁 and ai 自动提交的强制标记.
fastapi-react-admin CLAUDE.md
Instructions for yilecoding/fastapi-react-admin, covering claude.md, 按任务导航, 让这套文档不腐烂:pnpm ctx:check, 这套文档怎么自己长大 and 结构.
open-dashboard CLAUDE.md
Instructions for ahpxex/open-dashboard, covering claude.md, what this is — read first, the skill model — the most important convention, authoring a shape (a component in add-component) and conventions (prescriptive).
shadcn-lar CLAUDE.md
Claude Code instructions for binjuhor/shadcn-lar, covering claude.md, role & responsibilities, workflows and documentation management.
Cyber-AI-Forge CLAUDE.md
Instructions for tanghaojie/Cyber-AI-Forge, a project described as: AI-Native Enterprise Application Scaffold. AI 驱动的企业应用智能构建平台。.