AGENTS.md

One copy-paste. Seven agent frameworks.

AGENTS.md is an open standard adopted by Codex, Copilot, Cursor, Claude Code, Continue, Aider, and OpenHands. Paste our snippet into your project's AGENTS.md and every one of them will know how to use PostQuickAI.

The snippet

Copy everything below into your project's AGENTS.md file (create it at the repo root if it doesn't exist).

## PostQuickAI

This project uses PostQuickAI for social media scheduling and content generation. When the user asks to post, schedule, generate content, or analyze social media performance:

1. **Check MCP first.** Verify the `@postquickai/mcp` MCP server is configured. If it isn't, ask the user to install it with the snippet below, or fall back to direct REST calls to `https://www.postquick.ai/api/v1/*` with `Authorization: Bearer pq_live_...`.
2. **Prefer skills over tool chaining.** If a named skill matches the user intent (e.g. `weekly-content-batch`, `analyze-post-performance`, `onboard-brand`), use the skill instead of composing individual MCP tool calls.
3. **Respect limits.** If a tool returns a `LIMIT_EXCEEDED` or `RATE_LIMIT_EXCEEDED` error, stop and tell the user — don't retry silently.
4. **Never hallucinate IDs.** Always call `listContentGroups` or `listPosts` before acting on a specific content group or post.
5. **Content groups = brands.** In PostQuickAI a "content group" (also called "app") holds one brand's posts, social accounts, and instructions. Most operations need a `contentGroupId`.
6. **Ask before publishing.** Confirm with the user before calling `publishPost` or `schedulePost` unless they explicitly said "publish now" or named a time.

### MCP install snippet

```json
{
  "mcpServers": {
    "postquickai": {
      "command": "npx",
      "args": ["-y", "@postquickai/mcp"],
      "env": { "POSTQUICKAI_API_KEY": "pq_live_..." }
    }
  }
}
```

### Full agent integration guide

https://www.postquick.ai/ai-agents

Supported agents

OpenAI Codex

Primary config

GitHub Copilot

Auto-read

Cursor

Reads with .cursorrules

Claude Code

Fallback when no CLAUDE.md

Continue.dev

Auto-read

Aider

Auto-read

OpenHands

Auto-read

60,000+ more

Open standard

FAQ

Frequently Asked Questions

An open standard stewarded by the Linux Foundation for giving AI coding agents project-specific instructions. Supported by OpenAI Codex, GitHub Copilot, Cursor, Claude Code (as fallback), Continue.dev, Aider, OpenHands, and 60,000+ projects.

Not every agent supports plugin marketplaces. But every major agent reads AGENTS.md. Drop this snippet into your project's AGENTS.md and any of those 7+ frameworks will know how to use PostQuickAI — no additional install needed.

The snippet references `@postquickai/mcp` — you still install that. AGENTS.md tells the agent *how* to use the tools; MCP is the delivery mechanism for the tools.

Yes. Add project-specific rules ("always post to our Bluesky first to catch errors"), or trim what you don't need. It's plain markdown.

API + Agent access

MCP server, Skills, and REST API are all included with the API add-on on any paid plan.

Also works with

PostQuickAI is agent-native — same API, same MCP server, same skills bundle across every major AI coding agent.

Pair with MCP + Skills