Skip to content

Agent skills

Skills that turn an AI coding agent into a real teammate.

Three coherent skill packs — the Dailybot Agent Skill, Deep Work Plan, and AI Diff Reviewer — that install into every major AI CLI. Report progress. Complete check-ins. Draft and execute a gated plan. Review the branch before you push. All of it lands where your team already looks.

The point of a skill

  • One skill = one Dailybot verb

    Every action your agent takes maps 1:1 to a documented public endpoint. No hidden magic.

  • Every AI CLI, wired the same way

    Claude Code, Cursor, OpenAI Codex, Gemini, Copilot. One install per agent, same skill pack.

  • Every action lands in the standup

    Your agent stops being a black box. Its work shows up where the team already looks.

  • MIT-licensed, cross-agent by design

    You can fork it, strip our branding, and ship it. Nothing about it is locked to Dailybot.

The Dailybot skill pack in numbers

27

sub-skills across the three packs

5+

AI coding agents supported

MIT

licensed, end to end

1:1

mapped to public API verbs

Three skill packs. Twenty-seven verbs. Zero black boxes.

Not a marketplace. Not a plugin gallery. Three deeply-crafted skill packs — teammate verbs, gated multi-step plans, and a dual-surface code reviewer (Action + skill).

Dailybot Agent Skill

MIT

Every Dailybot verb your coding agent needs to be a real teammate.

One skill pack. Fourteen coordinated capabilities, each mapped to a Dailybot API verb. Install once and your agent can report progress, create and complete check-ins, send email, give and browse kudos, post on Slack/Teams/Discord/Google Chat, create and manage forms, open group conversations, ask the AI assistant, browse your org, and more — from inside the same coding session it already runs in.

Sub-skills · 13

  • report1–3 sentence standup updates. English only, WHAT + WHY.
  • checkinList, complete, create, and configure check-ins — schedule, questions, AI settings.
  • chatSend and edit messages on Slack, Teams, Discord, or Google Chat — including as your own identity.
  • kudosGive recognition to teammates or teams, and browse the kudos feed and wall of fame.
  • formsFull lifecycle plus authoring — create forms, configure workflow states, questions, and permissions.
  • teamsResolve teams, browse user profiles, and inspect org context.
  • healthAnnounce the agent online/offline to the team.
  • emailSend an email to a confirmed recipient.
  • conversationOpen a group DM with teammates and the Dailybot bot.
  • askAsk the Dailybot AI a question — headless, from the terminal.
  • messagesCheck for pending instructions and messages from your team.
  • workflowBrowse and inspect org-level automation workflows.
  • channelsDiscover available report channels for forms and check-ins.

Works with

Claude CodeCursorOpenAI CodexGoogle GeminiGitHub Copilot

Install

npx skills add DailybotHQ/agent-skill

Deep Work Plan

MIT

Turn any ambiguous goal into a numbered, gated plan any AI coding agent can execute.

The flagship agent methodology from Dailybot. DWP takes a fuzzy goal — "rebuild our developer portal", "migrate the payments service" — and produces a numbered, atomic, validated plan under `.dwp/plans/`. Each task has an explicit validation gate. The agent runs one task at a time and refuses to move on until validation passes.

Sub-skills · 8

  • createDecompose a goal into a numbered, gated plan.
  • executeRun the plan task-by-task, validating every gate.
  • refineAdd, remove, or reorder tasks without losing progress.
  • resumeReconstruct state and continue an interrupted plan.
  • statusRead-only progress report — completed, in-progress, blocked.
  • verifyCONFORMANT / NOT CONFORMANT audit against the plan spec.
  • onboardRe-run onboarding on an existing repo — reconciles, never clobbers.
  • authorAuthor or update a repo’s skills, agents, and commands.

Works with

Claude CodeCursorOpenAI CodexGoogle GeminiGitHub Copilot

Install

npx skills add DailybotHQ/deepworkplan-skill

AI Diff Reviewer

MIT

One review methodology for CI and your coding agent — Action + skill.

An LLM-driven code reviewer with two surfaces. Locally it installs as a coding-agent skill so you can review the current branch before you push. In CI it runs as a GitHub Action with inline comments and severity gating. Five capabilities — review, generate-extension, setup, open-pr, apply-review. Pin `@v2` on both sides for local ↔ CI parity.

Sub-skills · 5

  • reviewLocal review of the current branch — same methodology as CI.
  • generate-extensionAuthor `.review/extension.md` house rules shared by local + CI.
  • setupWizard to install the GitHub Action + reference for every input.
  • open-prDraft PR title + body from the diff and create/edit with gh.
  • apply-reviewWalk CI findings — apply, defer, or skip with per-finding consent.

Works with

Claude CodeCursorOpenAI CodexGoogle GeminiGitHub Copilot

Install

npx skills add DailybotHQ/ai-diff-reviewer --skill ai-diff-reviewer

What every Dailybot skill promises.

Every skill in the pack follows the same four rules. That’s the contract — with your agent, and with your team.

01

1:1 API mapping

Every sub-skill maps to a single public Dailybot endpoint. No new surfaces to learn — if it works via the API, it works via the skill.

02

Dual-auth parity

Reads either `dailybot login` or `$DAILYBOT_API_KEY`, chooses whichever is available. Personal OTP for developers, long-lived key for CI. Same access.

03

Per-repo profile

A `.dailybot/profile.json` per repo standardizes author name, default metadata, and reporting cadence — every contributor’s reports sign and tag consistently.

04

Standup-wired

Every action lands in the same daily feed your team already reads. Your agent’s work is visible, not buried in a private log.

One line to install. Every verb in your agent’s hands.

The `dailybot` CLI is the single binary every skill calls under the hood. Install it once and your agent — or any script, CI job, or shell — already has every Dailybot verb available.

Recommended

Install the Dailybot CLI

One binary. Every Dailybot verb. Works from any shell, any script, any CI job. This is what the Agent Skill actually calls under the hood — install this and your agent already has the full toolbox, no additional wiring required.

curl -fsSL https://cli.dailybot.com/install.sh | bash

Optional

Wire the skill directly into your AI CLI

If you already work inside an AI coding CLI, the Agent Skill can install into its skills folder with a single `npx skills add` command. The verbs still call the CLI you just installed — this only teaches your agent to reach for them.

npx skills add DailybotHQ/agent-skill

The road so far.

The public roadmap lives in each repo’s CHANGELOG. Here is the short version.

Past

Dual-auth parity, end to end

Both credentials — `dailybot login` and `DAILYBOT_API_KEY` — grant identical access. Parity is a written guarantee, not a footnote.

Now

27 sub-skills across three packs

Agent Skill (14 capabilities including env) + Deep Work Plan (8) + AI Diff Reviewer (review, generate-extension, setup, open-pr, apply-review) — all shipping today.

Next

More harnesses, tighter feedback

More supported agent frameworks, more sub-skills, and a tighter feedback loop with the community that ships against them every day.

Install once. Turn your agent into a teammate today.

The Agent Skill is one command in your AI CLI of choice. Deep Work Plan is a second. AI Diff Reviewer is a third — local review plus the GitHub Action. Everything else is public documentation.