기본 콘텐츠로 건너뛰기

Warp AI Playbook: Agent Mode for Daily Terminal Work

Summary: Warp is no longer “just a pretty terminal.” It is an agentic development environment: a modern terminal, a built-in Warp Agent that runs real shell commands with your approval, a Warp Agent CLI you can run in any terminal, and optional cloud agents / factories for unattended work. This playbook shows three daily workflows verified against Warp’s docs and pricing page on 2026-11-05 — start an agent conversation in the app with sensible permissions, run the same agent from the CLI, and reuse Drive prompts plus notifications — then helps you pick Free, Build, or Max without inventing limits.

What Warp is now (terminal + Warp Agent + cloud at a glance)

Warp’s documentation describes the product as an agentic development environment. The pieces that matter for day-to-day work are:

  • Warp terminal — a GPU-accelerated terminal for macOS, Windows, and Linux with block-based output, modern editing, and the surface where Agent Mode lives.
  • Warp Agent — Warp’s built-in coding agent. It runs multi-step tasks by executing commands in a real terminal, reading live output, and asking you before risky actions land. The same agent is available in the Warp app, via the Warp Agent CLI in any terminal or over SSH, and as a cloud agent on Warp’s Automation Platform.
  • Cloud agents and Factories (high level) — cloud agents run in the background from a trigger, schedule, or integration without keeping a local session open. Warp Factories sit on top of that platform when you want a team of cloud agents to triage, implement, review, and verify work. This playbook stays on local daily use; cloud and factories are the next layer once the local agent habits stick.

Official docs stress that context follows the agent: Rules, Skills, MCP servers, and Codebase Context apply the same way in the app, the CLI, and the cloud. You approve commands and review diffs before changes land, and you can dial autonomy up or down with Agent Profiles. See the official screenshots at the end of this post for the Profiles and Permissions UI.

Playbook 1: Daily agent work inside the Warp app

This is the highest-signal path if you already live in Warp as your terminal.

Install and open Warp

Platform support from the install guide: macOS (Intel and Apple silicon, 10.14+ with Metal), Windows 10 version 1903+ or Windows Server 2022+, and Linux with glibc >= 2.31 plus OpenGL ES 3.0+ or Vulkan.

  • macOS: download from warp.dev/download or brew install --cask warp.
  • Windows: download the installer or winget install Warp.Warp.
  • Linux: use the package for your distro from the download page (deb / rpm / pacman / AppImage and others).

Sign in with a Warp account. Agent Mode replaced the older AI chat panel; starting an AI query opens a conversation tied to your session.

Start a conversation and keep related work together

  1. Open a project directory in Warp (cd into a Git repo, or open the folder from the file tree). If Codebase Context is enabled, Warp detects the Git repo and begins indexing; check Settings > Code > Indexing and projects until status shows Synced. Free plan indexing supports up to 5,000 files per repository; larger repos need ignore files or a higher plan.
  2. Start a new Agent Mode conversation: on macOS press ⌘+Enter (or use /new); on Windows/Linux press Ctrl+Shift+Enter. You can also open the Conversations menu (⌘+Y / Ctrl+Shift+Y) and choose New Conversation.
  3. Ask in plain language about the repo, for example “Explain the architecture of this project” or “What are the main entry points?” When Codebase Context is synced, answers should cite real files rather than generic guesses.
  4. Keep follow-ups in the same conversation when the next question builds on the last answer. Start a fresh conversation for unrelated work — long threads get slower and noisier. Queue follow-ups while the agent is still working if you use prompt queueing.

Set permissions so “approve” means something

Go to Settings > Agents > Profiles. Each profile sets base model, autonomy, and tool access. Permission types include applying code diffs, reading files, creating plans, executing commands, interacting with running commands (Full Terminal Use), and asking clarifying questions.

Autonomy levels documented by Warp:

  • Agent Decides — acts when confident, asks when uncertain.
  • Always ask — explicit approval every time (good for sensitive actions).
  • Always allow — no confirmation for that action type.
  • Never — the agent will not take that action.

Two gotchas the permissions docs call out explicitly:

  1. The Command denylist beats “Always allow.” Default denylist patterns include things like curl, wget, rm, and eval. If the agent keeps asking about those, check the denylist before you crank every permission to Always allow.
  2. Run until completion (auto-approve for the current task: ⌘+Shift+I on macOS, Ctrl+Shift+I on Windows/Linux) bypasses your personal denylist by default. To keep the denylist during auto-approve, turn off Allow auto-approve to bypass command denylist under Settings > Agents > Warp Agent > Input. Team denylist rules from the Admin Panel are never bypassed.

A practical daily profile: Always ask (or Agent Decides) for execute/apply-diff on production repos; allowlist only read-only patterns such as ls, grep, find, which, echo if you want fewer prompts without going YOLO.

Playbook 2: Same agent from any terminal via Warp Agent CLI

When you are on a remote box, inside another terminal emulator, or you simply do not want the full Warp app open, the Warp Agent CLI runs the same agent with warp.

  1. Install (install script preferred because it auto-updates): macOS/Linux curl -fsSL https://app.warp.dev/download/agent-cli | bash; Windows PowerShell Invoke-RestMethod "https://app.warp.dev/download/agent-cli.ps1" | Invoke-Expression; or Homebrew brew install --cask warp-agent-cli (Homebrew does not auto-update — use brew upgrade --cask warp-agent-cli when prompted).
  2. Verify with warp --version.
  3. Run warp. The CLI shows a verification link and device code, opens the browser, and unlocks after you approve. Same Warp account as the app; the app itself is not required. For CI/headless use, prefer WARP_API_KEY=... over putting the key on the command line.
  4. Type a natural-language prompt such as What does this project do? and press Enter. Approve or reject permission requests the same way you would in the app.
  5. For a quick shell command without leaving the CLI, type ! on an empty input to enter Shell mode, run something like git status, then Backspace on empty input to return to agent prompts. Optional: /natural-language-detection classifies recognized shell vs natural language as you type.
  6. Exit with Ctrl+C twice or /exit. The CLI prints a resume command: warp --resume CONVERSATION_TOKEN.

Free plan includes Warp Agent CLI access according to the public pricing page. Usage for Warp-provided models still draws from your allowance or pay-as-you-go balance; BYOK routes inference to your provider instead.

Playbook 3: Reusable Drive prompts and agent notifications

Once the agent is part of your day, two features reduce friction without requiring cloud factories.

Save parameterized prompts in Warp Drive

Warp Drive stores Workflows, Notebooks, Prompts, and Environment Variables for personal use or a team. Open Drive from the status bar, the web experience, or the side panel (⌘+\ on macOS, Ctrl+Shift+\ on Windows/Linux).

  1. In Warp Drive, click + and choose Prompt.
  2. Name it, write the natural-language query, and add arguments with {{argument}} or “New argument” (names: letters, digits, hyphen, underscore; first character not a digit).
  3. Run it later from Drive, or search the Command Palette / Command Search (type prompts: to list prompts). Use Shift+Tab to cycle arguments when the prompt is selected.

Good candidates: “summarize git diff for a PR description,” “generate conventional commit message from staged changes,” “explain failing CI log pasted as context.” Keep secrets in Drive Environment Variables rather than hard-coding them into prompt text.

Notice when the agent needs you

Warp Agent notifications work out of the box: Complete, Request (blocked on approval/input), and Error. You get toast notifications, a notification mailbox (bell icon), tab status badges, and desktop alerts when Warp is in the background. Third-party CLI agents (Claude Code, Codex, OpenCode) need a one-time notification setup documented on their Warp pages — skip that until you intentionally run those harnesses inside Warp.

How to choose: Free vs Build vs Max

Prices and inclusions below are taken from warp.dev/pricing and Warp’s plans/billing docs as of 2026-11-05. Check your own Settings > Billing and usage for the allowance that actually applies to your seat — existing seats can differ from new-plan marketing numbers.

Plan Price (USD) Agent usage included Key limits / features (from pricing + docs)
Free $0 / month No bundled Warp Agent usage; pay-as-you-go for Warp-provided inference (purchases priced at a premium) or BYOK Modern terminal; Warp Agent CLI access; bring your own AI inference*; limited cloud agents; limited Warp Drive / collaboration; limited cloud conversation storage; Free codebase indexing up to 5,000 files per repository
Build $20 / month (monthly) or $18 / month when billed annually $20 of agent usage included each month at API rates; add-on usage, auto-reload, team-wide spend cap Everything in Free, plus full Warp Agent access (frontier and open-source models from OpenAI, Anthropic, z.ai, and more); extended cloud agents; highest codebase indexing limits; unlimited Warp Drive objects and collaboration; unlimited cloud conversation storage; private email support
Max $200 / month (monthly) or $180 / month when billed annually $200 of agent usage included each month (10× Build); add-on usage, auto-reload, team-wide spend cap Everything in Build, with the larger included usage pool

*Bring Your Own API Key (BYOK) is documented for Free and eligible paid plans for individuals and organizations with 10 or fewer employees (Anthropic, OpenAI, or Google keys). Larger orgs need Business/Enterprise with team-managed keys. ChatGPT consumer subscription linking is supported for eligible OpenAI models; Claude consumer subscriptions are not — use an Anthropic API key via BYOK instead.

Decision shortcuts: stay on Free if you mainly want the terminal, light agent trials, and BYOK. Move to Build when you want a predictable $20/month included pool at API rates plus unlimited Drive collaboration. Choose Max when you are burning through Build’s included usage every month and want the 10× pool without constantly buying add-ons. Team SSO and admin controls start at Business ($50/user/month monthly, $45 annual on the pricing page) — outside the personal Free/Build/Max choice.

Included usage resets monthly even on annual subscriptions and does not roll over. Purchased usage is separate and rolls over until its original 12-month expiration. Regular shell commands and non-AI terminal features do not consume agent usage; Generate and AI Autofill in Workflows do.

Editor’s note

Who this is for: developers who already live in the terminal and want an agent that runs real commands with visible approvals — not a separate chat tab you copy-paste into. Especially useful if you hop between laptop and remote SSH sessions (app + Agent CLI).

Who this is not for: people who only want a hosted no-code agent builder, or orgs that need SAML SSO and admin-managed BYOK on day one (look at Business/Enterprise instead of Free/Build/Max).

Gotchas verified in docs: (1) Denylist overrides Always allow — remove curl/rm patterns or expect prompts. (2) Run until completion bypasses personal denylist unless you disable that bypass. (3) Free has no bundled agent usage — BYOK or paid usage purchases required for sustained Warp-hosted inference. (4) Homebrew Agent CLI installs do not auto-update. (5) “Monthly request/credit limit exceeded” means check Billing and usage, then buy usage, enable auto-reload, switch to BYOK, or wait for the monthly reset.

Related on this blog: if you compare agentic editors and CLIs, see our Raycast AI playbook, Zed AI tutorial (agent panel + own keys), and Claude Code error fixes.

FAQ

Q1. Does the Free plan include Warp Agent CLI?
Yes. The public pricing page lists Warp Agent CLI access on Free. Free does not include bundled Warp Agent usage for Warp-provided models; use pay-as-you-go purchases (at a premium) or bring your own inference.

Q2. Can I bring my own LLM API key?
Yes. Warp documents BYOK for Anthropic, OpenAI, or Google keys on Free and eligible paid plans for individuals and organizations with 10 or fewer employees. Larger organizations need Business or Enterprise with team-managed keys/endpoints. Your provider bills that inference separately; Warp may still meter compute/platform for eligible cloud runs.

Q3. Why does the agent still ask before running curl or rm?
Those commands match the default Command denylist. The denylist takes precedence over Always allow and over Agent Decides. Edit the denylist under Settings > Agents > Profiles, or use Run until completion knowing it bypasses the personal denylist by default unless you change the Input setting.

Use Warp as your daily terminal first, then let Agent Mode take the repetitive investigation and patch loops under permissions you actually understand — that combination is the playbook, not a bigger subscription by default.

Last verified: 2026-11-05

Sources:

Image credit: official Warp docs screenshots (Agent Profiles, Permissions, toast notification) from docs.warp.dev Astro assets, plus an own screenshot of the public warp.dev/pricing page — captured 2026-11-05. See the official screenshots at the end of this post.

Uploaded Image Uploaded Image Uploaded Image Uploaded Image

댓글

이 블로그의 인기 게시물

브루(Vrew) 요금제 정리: 무료 200크레딧으로 어디까지 쓸 수 있나

요약: 영상 자막 편집기 브루(Vrew)는 2026년 4월 22일부터 기능별 사용량 대신 '통합 크레딧' 하나로 요금을 매겨요. 무료 플랜은 매달 200크레딧이 들어오는데, 음성 분석(자동 자막)만 쓰면 최대 200분 분량이라 짧은 영상 몇 편은 충분하지만, 영상 초반의 Vrew 마크를 지우려면 유료 플랜이 필요해요. 이 글은 브루 공식 가격 페이지와 고객센터 공지를 2026년 10월 23일에 다시 확인해, 크레딧 계산법과 플랜 고르는 기준을 정리했어요. 목차 브루 요금 구조: 통합 크레딧이란 1크레딧으로 할 수 있는 일 무료 플랜으로 되는 것과 안 되는 것 유료 플랜 4종 비교표 월 구독·연 구독·기간 이용권, 무엇이 이득일까 상황별 추천: 학생·직장인·유튜버 크레딧이 모자랄 때: 사용 기록 지우개 업그레이드·환불 전에 알아둘 규칙 운영자 메모, 자주 묻는 질문 1. 브루 요금 구조: 통합 크레딧이란 예전 브루는 플랜마다 음성 분석 몇 분, AI 목소리 몇 자, 번역 몇 자처럼 기능별 한도를 따로 줬어요. 그런데 보이저엑스(브루 운영사)가 실제 이용 데이터를 보니 한두 가지 기능만 집중해서 쓰는 사람이 훨씬 많았다고 해요. 그래서 2026년 4월 22일 업데이트부터는 플랜마다 정해진 크레딧을 받고, 어떤 기능에 쓸지는 내가 정하는 방식으로 바뀌었어요. 플랜별 매달 지급량은 Free 200 · Light 2,000 · Standard 10,000 · Business 12,000크레딧 이에요. 공식 공지에 따르면 무료 사용자는 매달 1일에, 유료 사용자는 매월 결제일에 충전되고, 남은 크레딧은 다음 달로 넘어가지 않아요. 바뀐 뒤 무엇이 좋아졌나 공지에 나온 예시가 이해하기 쉬워요. Standard 플랜에서 AI 목소리만 쓰면 예전에는 최대 50만 자였지만, 지금은 1만 크레딧을 모두 AI 목소리에 몰아 쓸 수 있어 최대 100만 자가 돼요. 반대로 여러 기능을 골고루 쓰는 사람은 크레딧이 기능끼리 ...

에이닷 전화 vs 익시오: 통화 녹음·AI 요약, 내 통신사·폰에 맞는 앱 고르기

요약: 통화 녹음과 AI 요약을 쓰고 싶다면 먼저 내 통신사와 휴대폰 종류부터 확인해야 해요. LG유플러스 고객은 무료 앱 익시오, SKT 고객은 에이닷 전화가 가장 기능이 많고, 다른 통신사의 아이폰 사용자는 iOS에 들어 있는 기본 통화 녹음·전사 기능이 현실적인 선택이에요. 이 글은 2026-10-30에 SKT·LG유플러스·Apple 공식 페이지와 도움말을 직접 확인해 세 가지를 비교했어요. 왜 '앱 기능'보다 '내 조건'을 먼저 봐야 할까요 통화 요약 앱 비교 글을 보면 기능 목록부터 나열하는 경우가 많아요. 그런데 실제로 써 보려는 사람에게 가장 먼저 걸리는 건 기능이 아니라 이용 조건 이에요. 익시오는 LG유플러스 고객 전용이고, 에이닷 전화는 누구나 설치할 수 있지만 통역콜·baro 같은 기능은 SKT 고객에게만 열려 있어요. 아이폰은 운영체제 구조상 앱이 통화를 마음대로 녹음하기 어려워서, 통신사 앱이 안 되면 iOS 기본 기능에 기대야 해요. 그래서 이 글은 "어떤 앱이 제일 좋은가"보다 "내 통신사·내 폰에서 무엇을 쓸 수 있나" 순서로 정리했어요. 기능 차이는 그다음이에요. 한눈에 비교: 에이닷 전화 vs 익시오 vs 아이폰 기본 통화 녹음 구분 에이닷 전화 (SKT) 익시오 ixi-O (LG유플러스) 아이폰 기본 통화 녹음 (Apple) 누가 쓸 수 있나 통신사 구분 없이 앱 설치 가능. 단, 통역콜·baro·그룹통화 발신 등은 SKT 고객 전용 U+ 모바일 고객. 타 통신사, 만 14세 미만, 선불요금제 고객은 제한 (공식 사이트 FAQ상 알뜰폰 중 KB리브엠은 예외) 지원 국가·언어의 iPhone 사용자. 1명과 통화할 때만 녹음 가능 지원 OS·기기 Android 10 이상, iOS 16 이상 iOS 17 이상 iPhone 12 이상 / Android...

타입캐스트 vs 클로바더빙: 수익 나는 채널에 쓸 때 무료·유료 조건 비교

요약 타입캐스트와 클로바더빙은 둘 다 글만 입력하면 한국어 AI 목소리를 만들어 주는 서비스지만, 무료로 쓸 수 있는 범위가 꽤 다릅니다. 클로바더빙 무료는 매달 15,000자까지 쓸 수 있는 대신 수익이 생기는 채널에는 쓸 수 없고, 타입캐스트 무료는 처음에 주는 약 5분 분량의 다운로드로 끝나는 체험판에 가깝습니다. 광고 수익이 붙은 유튜브나 쇼핑몰 영상에 넣을 목소리라면 처음부터 유료 플랜 조건을 보고 고르는 편이 안전합니다. 왜 이 두 서비스를 같이 비교하나 쇼츠 내레이션, 수업 자료 더빙, 상품 소개 영상처럼 "내 목소리 대신 AI 목소리"가 필요할 때 한국 사용자들이 가장 많이 찾는 곳이 네오사피엔스의 타입캐스트 와 네이버의 클로바더빙 입니다. 기능 소개만 보면 비슷해 보이지만, 실제로 문제가 되는 건 음질보다 "이 목소리를 내 채널에 써도 되나" 입니다. 그래서 이 글은 기능 나열보다 무료 사용 범위, 출처 표기 의무, 유료 플랜 가격과 다운로드 한도를 중심으로 정리했습니다. 모든 수치는 2026년 10월 25일에 두 서비스의 공식 요금 페이지와 고객센터 도움말에서 직접 다시 확인한 내용입니다. 한눈에 보는 비교표 항목 타입캐스트 무료 타입캐스트 베이직 클로바더빙 Free 클로바더빙 Standard 월 요금 0원 9,900원 (연간 결제 시 10% 할인) 0원 19,900원 (네이버 클라우드 플랫폼에서 결제) 무료·포함 사용량 음성 생성·재생 무제한, 다운로드는 최초 3,000크레딧(약 5분) 1회 지급 월 30,000크레딧(약 60분) 월 15,000자, 다운로드 월 20회, 프로젝트 누적 5개 (매월 1일 갱신) 월 30,000자, 다운로드 월 40회, 프로젝트 누적 20개 쓸 수 있...