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/mdcms-ai/mdcms/agents-mdgit clone --depth 1 https://github.com/mdcms-ai/mdcmsWhat 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 | $0.03337 | $0.03337 |
| Opus 5 | $0.01669 | $0.01669 |
| Sonnet 5 | $0.00667 | $0.00667 |
| Haiku 4.5 | $0.00334 | $0.00334 |
Grade A, and why
mdcms 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 2d 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.
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 — 241 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MDCMS
Vision
MDCMS is a Markdown-first headless CMS where the database holds the content but developers work with local files. You pull content down, edit with whatever tools you like, and push it back. The editing experience feels like a filesystem, without the problems that come with actually storing content in one.
Three interfaces share the same data layer: a CLI for developers, an embeddable Studio for editors, and a REST API that AI agents and applications consume directly. The goal is that nobody blocks anyone else. An editor publishing a page and an agent rewriting 500 posts at once go through the same validation, the same permissions, and the same version history.
Direction
The module system is how MDCMS gets extended. Server actions, CLI commands, and Studio UI surfaces all hook in through the same module contract. Community contributions should follow this pattern rather than patching core code.
Upcoming work is focused on live preview (real-time content rendering in the consumer frontend), MCP integration (Model Context Protocol for agent-driven content operations), multiple spaces (team-scoped content organization), and real-time collaboration (multi-user editing via CRDTs).
Repository layout
| Path | What it is |
|---|---|
apps/server |
Elysia HTTP server, the backend for everything |
apps/cli |
CLI binary (mdcms) for push/pull/sync workflows |
apps/studio-example |
Next.js app that embeds the Studio component |
apps/studio-review |
Internal contract-consumer for preview mocks; kept in sync with contracts |
packages/shared |
Contracts, types, and schema utilities used everywhere |
packages/sdk |
Client SDK for reading content from applications |
packages/studio |
The embeddable React Studio component |
packages/modules |
First-party module registry |
skills/ |
Public MDCMS skills pack (skills.sh-distributable) for users adopting MDCMS in their projects |
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.
- 2d ago First seen · 241 lines · 3,337 tokens per session scan A fbb6708ddcdb
mdcms AGENTS.md is an instructions file published in the GitHub repository mdcms-ai/mdcms (22 stars, last pushed 4d ago), licensed MIT. It adds 3,337 tokens to every session, about $0.0167 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
payload CLAUDE.md
Instructions for payloadcms/payload, covering claude.md, project structure, key directories, architecture notes and quick start.
payload AGENTS.md
Instructions for payloadcms/payload, a project described as: Payload is the open-source, fullstack Next.js framework, giving you instant backend superpowers. Get a full TypeScript backend and admin panel instantly. Use Payload as a headless CMS or for building powerful applications.
intlayer CLAUDE.md
Instructions for aymericzip/intlayer, covering repo layout, commands (run from root), content declaration flow, git and commit conventions.
microfeed AGENTS.md
AGENTS.md instructions for microfeed/microfeed, covering repository guidance, development workflow, instance management and cloudflare deployment, source architecture and theme repositories.
sonicjs CLAUDE.md
Instructions for SonicJs-Org/sonicjs, covering sonicjs ai development guidelines, core technology stack, workspace boundary (conductor), architecture direction: document model (authoritative) and the 5 document tables (migration 0002documents.sql).
flowwink CLAUDE.md
Instructions for magnusfroste/flowwink, covering claude.md, commands, development (auto-runs migrations before starting), production build and lint.