dailybot

Dailybot Plugin for Claude Code

Connect Claude Code to your team via Dailybot. Report progress, check messages, send emails, and announce agent status — all integrated into your coding workflow.

What it does

This plugin gives Claude Code five capabilities for team collaboration through Dailybot:

Progress Reporting — When you complete meaningful work (shipping a feature, fixing a bug, finishing a task), the plugin sends a standup-style progress report to Dailybot. Reports are written in a human-first style: they describe what was accomplished and why it matters, not which files changed.

Team Messages — Check for pending messages and instructions from your team. Messages represent tasks, priorities, context, or feedback that should influence your current work.

Email — Send emails to anyone through Dailybot. Useful for notifications, summaries, follow-ups, or weekly reports. Replies are delivered back as messages to your agent inbox.

Health Status — Announce your agent's status (online, working, offline, degraded) so the team knows what's happening. Health checks also deliver any pending messages from the team.

Guided Setup — First-time setup walks you through CLI installation, authentication, and agent profile configuration step by step.

Install

From inside Claude Code, register the DailyBot marketplace once and install the plugin:

/plugin marketplace add DailyBotHQ/dailybot-claude-plugin
/plugin install dailybot@dailybot

The first command tells Claude Code to trust this repository as a plugin source. The second installs the dailybot plugin from that marketplace. From then on you can update with /plugin update dailybot@dailybot.

Or test locally during development without registering a marketplace:

claude --plugin-dir ./path/to/dailybot-claude-plugin

Setup (one time)

1. Dailybot CLI

The plugin requires the Dailybot CLI. Run /dailybot:setup for guided setup, or install it yourself:

# Option A: pip
pip install dailybot-cli

# Option B: install script
curl -sSL https://cli.dailybot.com/install.sh | bash

Requires Python 3.9+.

2. Authentication

Claude will guide you through login on first use. Or do it yourself:

dailybot login

This sends a verification code to your email. Any Dailybot team member can log in — you don't need to be an admin.

Don't have a Dailybot account yet? No problem — Claude can create a new Dailybot organization for you right from the terminal. Just tell Claude you don't have an account and it will handle the setup. You'll get a link to share with your team so they can connect Dailybot to Slack, Teams, Discord, or Google Chat.

3. Agent name (optional)

When you enable the plugin, Claude Code asks for an agent name. This is how your reports appear in Dailybot. Default is claude-code. You can change it to anything descriptive like my-backend-agent or deploy-bot.

Usage

Slash commands

Command	What it does
/dailybot:report	Send a progress report to your team
/dailybot:messages	Check for pending messages from your team
/dailybot:email	Send an email through Dailybot
/dailybot:health	Announce agent status and pick up messages
/dailybot:setup	Guided first-time setup (CLI install, auth, agent profile)

Natural language

You can also use natural language:

• "Send an update to Dailybot about what we did"
• "Report this to my team"
• "Do I have any messages?"
• "What should I work on next?"
• "Email this summary to [email protected]"
• "Go online" / "Check in with the team"

Automatic behavior

• Progress reporting: Claude detects when you've completed significant work and sends a report automatically. Trivial changes (typo fixes, lockfile updates, formatting) are skipped.
• Stop gate: When a turn completes with significant unreported work, Claude is reminded to offer a Dailybot report. The reminder is rate-limited (at most once per 60 seconds) and suppressed once a report is sent.
• Message check: At the start of each session, pending messages from your team are fetched automatically.

Hook architecture

The plugin uses a layered hook design to keep things invisible during normal use and only surface a nudge when there's real work to report.