v0.1.0

ddYour daily-driver CLI for AI-powered development

Hook into your project to give coding agents full context. Run your build-ship-monitor loop. Scaffold production Next.js apps when you need a new one.

$ curl -fsSL https://devdigest.dev/install.sh | sh

Value

Why developers use dd

One command, any agent

dd auto-detects Claude Code, Codex, or OpenCode on your system. No config needed. Just type dd and your prompt.

Full project context

dd reads your stack, config, and health before passing to the agent. Your AI gets the full picture, not just your prompt.

Ship faster

dd ship runs lint, test, build, deploy, and tag in one command. dd pr generates PR descriptions from your diff.

Workflow

The daily loop

Four steps. Every day. Start your session, build features, ship them, and monitor the results.

1

Start

dd next

dashboard: git, TODOs, dirty files, PRs

dd recap

recent git activity and stats

2

Build

dd "add billing to the dashboard"

natural language → best agent

dd diff

what changed?

dd test

auto-detect and run tests

dd test --pipe claude

run tests, auto-pipe failures to Claude

3

Ship

dd ship

lint → test → build → deploy → tag

dd pr

create PR with auto-generated description

dd changelog

generate changelog from commits

4

Monitor

dd report

open PRs, issues, CI status

dd latest

recently merged PRs

dd todo

scan for TODO/FIXME/HACK

Commands

Everything you need

32 commands across 7 categories. Every command supports -o json, --no-input, and --pipe claude.

Build

Pipe to agents
dd "prompt"

Auto-route to best installed agent with full project context

dd claude [prompt]

Pipe to Claude Code (with workflows: setup, feature, review, deploy)

dd codex [prompt]

Pipe to Codex

dd opencode [prompt]

Pipe to OpenCode

dd context

Generate project context (pipe to any agent or clipboard)

Run

Develop and test
dd dev

Start all services (Next.js + Convex) with prefixed output

dd test

Auto-detect framework (vitest, jest, playwright) and run tests

dd diff

Smart diff summary — files changed, stats, visual bars

Ship

Deploy and release
dd ship

Lint, test, build, deploy to Vercel, git tag (--dry-run to preview)

dd pr

Create GitHub PR with auto-generated title and description

dd changelog

Generate changelog from git commits between tags

dd clean

Remove build artifacts and caches

Monitor

Stay on top of your project
dd next

What's next? Dashboard of git, TODOs, dirty files, recent files, PRs

dd report

GitHub activity — open PRs, recent issues, CI status

dd latest

Recently merged PRs — catch up on changes

dd recap

Recent git activity — commits, stats, dirty state

dd todo

Scan codebase for TODO/FIXME/HACK/BUG annotations

dd status

Project dashboard — stack, health, agents, rules

dd doctor

Health check — env vars, deps, config

dd env

Env var status — what's set, what's missing

Knowledge

Research and rules
dd docs [target]

Fetch documentation as pipeable markdown

dd research [pkg]

Research any npm package (registry + docs)

dd rules add <names>

Install agent rules to .claude/rules/

Scaffold

New projects and integrations
dd init [name]

Create a new project with the DevDigest stack

dd add [integrations]

Add integrations to an existing project

dd generate <type> <name>

Generate page, API route, layout, or component

dd upgrade

Check for dependency updates (--apply to install)

Configure

Preferences and workflows
dd config show

Show project conventions

dd config global

Global preferences — preferred agent, fallbacks

dd workflow add <name>

Create custom reusable prompt workflows

dd agent add <names>

Install slash commands (reviewer, tester, debugger, deployer)

dd open [target]

Open integration dashboard in browser

dd update

Self-update to latest version (--force to reinstall)

Multi-Agent

Auto-detects your agent

dd finds Claude Code, Codex, or OpenCode on your system, injects your full project context, and pipes your prompt.

How it works

1

You type a prompt

dd "add billing to the dashboard"
2

dd gathers context

Stack, health, config, rules, dirty files, TODOs

3

Routes to best agent

Claude Code, Codex, or OpenCode — whichever is installed

4

Agent executes with context

Full project awareness, not a cold start

Agent-first design

Built for AI agents to call, not just humans.

-o json on every command

Agents never parse human text

--dry-run

Preview side effects before committing

--no-input

Fail instead of prompting (no hung agents in CI)

--pipe claude

Auto-pipe output to an agent without interactive menus

dd schema

Full CLI API as JSON in one call

SKILL.md

Structured manifest agents read to understand the tool

MCP server

23 tools exposed as an MCP server. Drop this into your config:

{
  "mcpServers": {
    "devdigest": {
      "command": "dd",
      "args": ["mcp"]
    }
  }
}

Set your preferred agent

$ dd config global set agent claude        # prefer Claude Code
$ dd config global set fallbacks codex,claude  # fallback order
$ dd config global set yes true            # auto-approve permissions

Get started

Install dd

One command. No dependencies. Self-updating.

Recommended (self-updating)

$ curl -fsSL https://devdigest.dev/install.sh | sh

Or via npm

$ npm install -g devdigest

Update

$ dd update
$ cd your-project
$ dd                                   # what's next? (auto-dashboard)
$ dd doctor                            # check project health
$ dd "add a settings page"             # auto-routes to best agent