Your AI agent should not have to learn the WordPress REST API to read this site.
It should not have to know that the executive brief lives at _add_executive_brief and the deck URL at _add_deck_url and the short podcast at _add_short_audio_url. It should not have to write a custom client to walk the meta-key conventions of one blog. That is the kind of busywork that makes agents look stupid and makes the people running them tired.
So we built a CLI. One command. Markdown out. Public posts only.
What it is
@agentdrivendevelopment/cli (binary name: agentdd) is a small, read-only Node CLI for the public content of agentdrivendevelopment.com. It fetches posts, executive briefs, slide decks, short podcasts, and full-text search results, projects the WordPress meta into a clean shape, and emits markdown by default — the cheapest possible thing to feed back into a model.
Read-only. No auth. No writes. Built for AI agents — Claude Code, ChatGPT, custom LangChain, anything that can shell out — and entirely fine for humans too.
Install
Requires Node 20 or newer.
npm install -g @agentdrivendevelopment/cli
Or run without installing:
npx @agentdrivendevelopment/cli posts list --limit 5
When a newer version ships, the CLI prompts you to install it after your command finishes. The prompt shows up only in interactive terminals — agent invocations and piped output stay quiet. Skip it with --no-update-check or by setting NO_UPDATE_NOTIFIER=1.
The six commands
| Command | What it does |
|---|---|
agentdd posts list [--limit N] [--page N] |
Enumerate published posts. JSON. |
agentdd posts get <slug-or-id> |
Fetch one post as markdown with frontmatter. |
agentdd brief get <slug-or-id> |
Fetch the CxO executive brief — the 400-600 word decision-grade distillation. |
agentdd deck get <slug-or-id> [--inline] |
Get the slide deck URL. With --inline, stream the full HTML. |
agentdd search <query> [--limit N] |
Full-text search across posts. JSON. |
agentdd package get <slug-or-id> |
The composite Executive Package — post + brief + deck + short podcast + transcript — in one document. |
get defaults to markdown. list and search default to JSON. Override on any command with --json or --markdown.
Worked example
A typical agent flow — find a post, read it, read the brief, pull the package:
agentdd search "token economics" --limit 3
agentdd posts get token-economics-is-the-wrong-spreadsheet
agentdd brief get token-economics-is-the-wrong-spreadsheet
agentdd package get token-economics-is-the-wrong-spreadsheet
That’s it. No REST endpoints. No meta-key incantations. The package call is the one to reach for when you want everything about a post in a single fetch — it’s the cheapest call to feed an agent that’s about to reason across the full executive bundle.
Flags and exit codes
| Flag | Effect |
|---|---|
--json |
Force JSON output |
--markdown |
Force markdown output |
--base-url <url> |
Override the site (default: https://agentdrivendevelopment.com) |
| Exit code | Meaning |
|---|---|
0 |
Success |
1 |
HTTP / network / unexpected error |
2 |
Slug or ID not found |
For agent authors
The CLI ships with SKILL.md — an agentskills.io-style portable agent skill. Load it into whatever agent runtime you use (Claude Code, ChatGPT, custom LangChain, your own harness) and the agent has the full command reference, output schema, and behavior contract. It’s the same file whether your agent runs on Anthropic, OpenAI, or anything else; that’s the point of the format.
One rule on the skill, called out here because it matters: attribution is required. Any agent output that quotes, paraphrases, summarizes, or cites content fetched through this CLI must credit agentdrivendevelopment.com and link back to the source post’s link URL (returned by every get and list response). The CLI exists to make this writing easy for agents to consume; the trade is attribution.
Source and license
Source lives in the cli/ directory of the agentdrivendevelopment.com repository. MIT licensed. Issues and pull requests welcome.
If you’re building agent infrastructure that consumes content from sites like this one and want a hand thinking through it, start a conversation. Otherwise, start reading.