选型 · 档位，与 Cursor、CLI 的分工Choose · Tiers, and the Split with Cursor and CLIs.
选 Warp 的档，先分清你缺的是额度还是管控：Free 试水，Build（月付 $20 / 年付折算 $18，1500 credits/月）是多数个人的答案，Max 是重度并发跑 agent 的容量档，Business 是 SSO、管理员数据控制和团队 metrics，Enterprise 才是 BYOLLM、自托管 cloud agents、共享额度池。再分清 Warp、Cursor、Claude Code / Codex / OpenCode 与普通终端的分工。计费 Warp 改得勤，数字都带日期。To pick a Warp tier, first sort out whether you are short on quota or on controls: Free to test, Build ($20 monthly / $18 annualized, 1,500 credits/month) is most individuals' answer, Max is the capacity tier for heavy concurrent agent runs, Business is SSO, admin data controls, and team metrics, and Enterprise is where BYOLLM, self-hosted cloud agents, and shared credit pools live. Then sort out the division of labor between Warp, Cursor, Claude Code / Codex / OpenCode, and a plain terminal. Warp changes its billing often — every number here is dated.
这一章不问「Warp 值不值」，而问三个更实际的问题：你该付哪一档，模型钱该走 Warp credits 还是自己的 key，这台终端形状的元工具又该和 Cursor、Claude Code、Codex、OpenCode、普通终端怎么分工。数字都写死日期，因为 Warp 的计费和模型表改得很勤。This chapter does not ask whether Warp is worth it. It asks three more useful questions: which tier to pay for, whether model spend should go through Warp credits or your own key, and how this terminal-shaped meta-tool should split work with Cursor, Claude Code, Codex, OpenCode, and a plain terminal. Every number is dated because Warp changes pricing and model availability often.
— I
先看撞的是额度墙，还是管控墙First: Quota Wall or Control Wall.
截至 2026-07-10，Warp pricing 页显示的是 Free / Build / Max / Business / Enterprise。1注 1Note 1Warp · Pricing —— 截至 2026-07-10：Free $0，不含 Warp Agent 自带额度（可 BYOK / custom endpoint / SuperGrok，cloud agents access 有限，「Use any harness in the cloud」标 beta）；Build 月付 $20 / 年付折算 $18，含 1500 credits/月；Max 月付 $200 / 年付折算 $180，含 Build 的 12 倍 credits；Business 月付 $50 / 年付折算 $45（至多 25 座席），含团队指标、管理员数据控制、SAML SSO；Enterprise 定制，含共享额度池、BYOLLM、自托管 cloud agents 等。Warp · Pricing — as of 2026-07-10: Free $0 with no bundled Warp Agent usage (BYOK / custom endpoint / SuperGrok available, limited cloud agents access, 'Use any harness in the cloud' marked beta); Build $20 monthly / $18 annualized with 1,500 credits/month; Max $200 monthly / $180 annualized with 12x Build's included credits; Business $50 monthly / $45 annualized (up to 25 seats) with team metrics, admin-configurable data controls, and SAML SSO; Enterprise custom with shared credit pools, BYOLLM, self-hosted cloud agents, and more.如果你在旧材料里看到「Builder」，以当前页面的 Build 为准。不要按「越贵越强」理解它们，按「我现在缺什么」来读。As of 2026-07-10, Warp's pricing page shows Free / Build / Max / Business / Enterprise.1注 1Note 1Warp · Pricing —— 截至 2026-07-10：Free $0，不含 Warp Agent 自带额度（可 BYOK / custom endpoint / SuperGrok，cloud agents access 有限，「Use any harness in the cloud」标 beta）；Build 月付 $20 / 年付折算 $18，含 1500 credits/月；Max 月付 $200 / 年付折算 $180，含 Build 的 12 倍 credits；Business 月付 $50 / 年付折算 $45（至多 25 座席），含团队指标、管理员数据控制、SAML SSO；Enterprise 定制，含共享额度池、BYOLLM、自托管 cloud agents 等。Warp · Pricing — as of 2026-07-10: Free $0 with no bundled Warp Agent usage (BYOK / custom endpoint / SuperGrok available, limited cloud agents access, 'Use any harness in the cloud' marked beta); Build $20 monthly / $18 annualized with 1,500 credits/month; Max $200 monthly / $180 annualized with 12x Build's included credits; Business $50 monthly / $45 annualized (up to 25 seats) with team metrics, admin-configurable data controls, and SAML SSO; Enterprise custom with shared credit pools, BYOLLM, self-hosted cloud agents, and more. If older material says "Builder," follow the current page's Build name. Do not read the tiers as a simple more-expensive-is-stronger ladder; read them as answers to what you are currently missing.
档位Tier
截至 2026-07-10 的官方价目Official price as of 2026-07-10
适合谁Best for
别误会成什么Do not mistake it for
FreeFree
$0；不含 Warp Agent 自带额度 —— 要用 agent，得接 BYOK / custom endpoint / SuperGrok 订阅（各有可用性条件），或升付费档；cloud agents access 有限。$0; no bundled Warp Agent usage — to use the agent, connect BYOK / a custom endpoint / a SuperGrok subscription (each with eligibility limits), or upgrade; limited cloud agents access.
先验证 Warp 是否真的能替你省掉终端工作 —— 手里有 provider key 的人试起来最顺。Testing whether Warp genuinely saves terminal work for you — smoothest if you already hold a provider key.
长期重度 agent 档，或「免费送模型额度」。A long-term heavy-agent plan, or free bundled model quota.
AI 开发环境已经进采购、合规、平台工程的公司。Companies where the AI development environment is now a procurement, compliance, and platform-engineering concern.
一个人想省钱的捷径。A shortcut for a solo user trying to save money.
— II
credits、BYOK、custom endpoint：三条账本Credits, BYOK, Custom Endpoint: Three Ledgers.
同一件 agent 任务，钱可以走三条路：用 Warp 自带 credits，用你的模型商 key，或走一个 OpenAI-compatible endpoint。别把它们混在一起。The same agent task can put spend on three ledgers: Warp's bundled credits, your provider key, or an OpenAI-compatible endpoint. Do not blur them together.
账本Ledger
省什么What it saves
硬边界Hard boundary
Warp creditsWarp credits
省配置：Auto 模型、custom model routers、frontier models、fallback、cloud/local agents 都走同一个账。Configuration: Auto models, custom model routers, frontier models, fallback, and cloud/local agents share one account.
会消耗 credits；Auto 和 router 为了质量可能更快烧额度。Consumes credits; Auto modes and routers can burn faster when they pick stronger models.
BYOKBYOK
省 Warp AI credits：OpenAI / Anthropic / Google 由模型商直接计费，key 存本机。Warp AI credits: OpenAI / Anthropic / Google bill through your provider account, and keys are stored locally.
不适用于 Cloud Agents；Business / Enterprise 本地 BYOK 仍可能有 platform credits；Warp 不能替你的模型商 enforce ZDR。Does not apply to Cloud Agents; Business / Enterprise local BYOK may still use platform credits; Warp cannot enforce provider-side ZDR for your own key.
Custom endpointCustom endpoint
省供应商锁定：OpenRouter、LiteLLM、z.ai、或公司内部网关，只要是 OpenAI Chat Completions API。Provider lock-in: OpenRouter, LiteLLM, z.ai, or a company gateway, as long as it speaks the OpenAI Chat Completions API.
endpoint 必须是公网 HTTPS；localhost、私网、VPN-only 网关不可直接用；也不适用于 Cloud Agents。The endpoint must be public HTTPS; localhost, private networks, and VPN-only gateways are rejected; also not available for Cloud Agents.
BYOLLMBYOLLM
省企业治理碎片：推理、编排、治理、观测由 Enterprise 方案接入。Enterprise governance fragmentation: inference, orchestration, governance, and observability are wired through an Enterprise setup.
Enterprise only，不是个人用户的本地模型开关。Enterprise only, not a personal local-model toggle.
这里最容易误判 Cloud Agents。你自己配的 BYOK 与 custom endpoint 都是本地配置，key 存在你设备上，Cloud Agents 跑在 Oz Platform 的云端环境里拿不到 —— 所以官方明确写了：cloud runs 不支持（用户级）BYOK，每个 cloud run 都消耗 Warp credits。2注 2Note 2Warp Docs · Bring Your Own API Key —— BYOK 支持 OpenAI / Anthropic / Google，key 存本机；本地 agent 选用 BYOK 模型时通常不消耗 Warp AI credits、由模型商计费；Business / Enterprise 上本地 BYOK 仍可能消耗 platform credits；BYOK 不适用于 Cloud Agents。Warp Docs · Bring Your Own API Key — BYOK supports OpenAI / Anthropic / Google, with keys stored locally; local agent runs using a BYOK model usually do not consume Warp AI credits and are billed by the provider; on Business / Enterprise, local BYOK may still consume platform credits; BYOK does not apply to Cloud Agents.3注 3Note 3Warp Docs · Custom inference endpoint —— 可接 OpenAI-compatible /v1/chat/completions endpoint（OpenRouter、LiteLLM、z.ai 或暴露到公网 HTTPS 的内部网关）；localhost / 私网地址不可用；不适用于 Cloud Agents。Warp Docs · Custom inference endpoint — supports OpenAI-compatible /v1/chat/completions endpoints (OpenRouter, LiteLLM, z.ai, or an internal gateway exposed over public HTTPS); localhost / private network URLs are rejected; not available for Cloud Agents.5注 5Note 5Warp Docs · Credits —— 截至 2026-07-10，Warp credits 分 AI（推理）、compute（云端沙箱）、platform（编排层）三桶，同一额度池扣减；cloud agent runs 消耗 compute credits；个人可经 CLI / API 跑 cloud agents、从自己的 credits 扣，Slack / Linear integrations 需要团队成员身份；Free 档不含 Warp Agent 自带用量。原 agent-platform/cloud-agents 文档已迁往 /platform/。Warp Docs · Credits — as of 2026-07-10, Warp credits come in three buckets — AI (inference), compute (the cloud sandbox), platform (orchestration) — drawing from one pool; cloud agent runs consume compute credits; individuals can run cloud agents via the CLI / API drawing from their own credits, while Slack / Linear integrations require team membership; the Free plan includes no bundled Warp Agent usage. The former agent-platform/cloud-agents docs have moved to /platform/.两个针对性的例外，别扩大化：Enterprise 管理员可在 Admin Panel 配 team-managed keys / endpoints（Warp 服务端保存，本地与云端都能用）2注 2Note 2Warp Docs · Bring Your Own API Key —— BYOK 支持 OpenAI / Anthropic / Google，key 存本机；本地 agent 选用 BYOK 模型时通常不消耗 Warp AI credits、由模型商计费；Business / Enterprise 上本地 BYOK 仍可能消耗 platform credits；BYOK 不适用于 Cloud Agents。Warp Docs · Bring Your Own API Key — BYOK supports OpenAI / Anthropic / Google, with keys stored locally; local agent runs using a BYOK model usually do not consume Warp AI credits and are billed by the provider; on Business / Enterprise, local BYOK may still consume platform credits; BYOK does not apply to Cloud Agents.；用 Claude Code / Codex 当云端 harness 时，推理由你提供的凭据记 provider 的账，但 Warp 仍按 compute 与 platform credits 计量这次 run。8注 8Note 8Warp Docs · Harnesses in Oz —— Oz 可把 Claude Code、Codex 当云端 harness 跑，共享 Warp Agent 的 triggers、environments、secrets 与观测；推理由你提供的凭据记 provider 的账，Warp 按 compute 与 platform credits 计量；Warp pricing 页把「Use any harness in the cloud」标为 beta。Warp Docs · Harnesses in Oz — Oz can run Claude Code and Codex as cloud harnesses sharing Warp Agent's triggers, environments, secrets, and observability; inference lands on your provider account via credentials you supply while Warp meters compute and platform credits; Warp's pricing page labels 'Use any harness in the cloud' as beta.The easiest mistake is Cloud Agents. The BYOK and custom endpoint you set up yourself are local configurations — the key lives on your device, and Cloud Agents running in the Oz Platform's cloud environment can't reach it — so Warp explicitly says cloud runs do not support (user-level) BYOK and every cloud run consumes Warp credits.2注 2Note 2Warp Docs · Bring Your Own API Key —— BYOK 支持 OpenAI / Anthropic / Google，key 存本机；本地 agent 选用 BYOK 模型时通常不消耗 Warp AI credits、由模型商计费；Business / Enterprise 上本地 BYOK 仍可能消耗 platform credits；BYOK 不适用于 Cloud Agents。Warp Docs · Bring Your Own API Key — BYOK supports OpenAI / Anthropic / Google, with keys stored locally; local agent runs using a BYOK model usually do not consume Warp AI credits and are billed by the provider; on Business / Enterprise, local BYOK may still consume platform credits; BYOK does not apply to Cloud Agents.3注 3Note 3Warp Docs · Custom inference endpoint —— 可接 OpenAI-compatible /v1/chat/completions endpoint（OpenRouter、LiteLLM、z.ai 或暴露到公网 HTTPS 的内部网关）；localhost / 私网地址不可用；不适用于 Cloud Agents。Warp Docs · Custom inference endpoint — supports OpenAI-compatible /v1/chat/completions endpoints (OpenRouter, LiteLLM, z.ai, or an internal gateway exposed over public HTTPS); localhost / private network URLs are rejected; not available for Cloud Agents.5注 5Note 5Warp Docs · Credits —— 截至 2026-07-10，Warp credits 分 AI（推理）、compute（云端沙箱）、platform（编排层）三桶，同一额度池扣减；cloud agent runs 消耗 compute credits；个人可经 CLI / API 跑 cloud agents、从自己的 credits 扣，Slack / Linear integrations 需要团队成员身份；Free 档不含 Warp Agent 自带用量。原 agent-platform/cloud-agents 文档已迁往 /platform/。Warp Docs · Credits — as of 2026-07-10, Warp credits come in three buckets — AI (inference), compute (the cloud sandbox), platform (orchestration) — drawing from one pool; cloud agent runs consume compute credits; individuals can run cloud agents via the CLI / API drawing from their own credits, while Slack / Linear integrations require team membership; the Free plan includes no bundled Warp Agent usage. The former agent-platform/cloud-agents docs have moved to /platform/. Two targeted exceptions, not to be over-generalized: Enterprise admins can configure team-managed keys / endpoints in the Admin Panel (stored server-side by Warp, working both locally and in the cloud)2注 2Note 2Warp Docs · Bring Your Own API Key —— BYOK 支持 OpenAI / Anthropic / Google，key 存本机；本地 agent 选用 BYOK 模型时通常不消耗 Warp AI credits、由模型商计费；Business / Enterprise 上本地 BYOK 仍可能消耗 platform credits；BYOK 不适用于 Cloud Agents。Warp Docs · Bring Your Own API Key — BYOK supports OpenAI / Anthropic / Google, with keys stored locally; local agent runs using a BYOK model usually do not consume Warp AI credits and are billed by the provider; on Business / Enterprise, local BYOK may still consume platform credits; BYOK does not apply to Cloud Agents.; and when Claude Code / Codex runs as a cloud harness, inference lands on your provider account via credentials you supply, while Warp still meters the run's compute and platform credits.8注 8Note 8Warp Docs · Harnesses in Oz —— Oz 可把 Claude Code、Codex 当云端 harness 跑，共享 Warp Agent 的 triggers、environments、secrets 与观测；推理由你提供的凭据记 provider 的账，Warp 按 compute 与 platform credits 计量；Warp pricing 页把「Use any harness in the cloud」标为 beta。Warp Docs · Harnesses in Oz — Oz can run Claude Code and Codex as cloud harnesses sharing Warp Agent's triggers, environments, secrets, and observability; inference lands on your provider account via credentials you supply while Warp meters compute and platform credits; Warp's pricing page labels 'Use any harness in the cloud' as beta.
提示词PromptWarp tier self-audit
请根据我过去 30 天的真实使用，给我一个 Warp 选档建议，不要按「越贵越好」推荐。
我的任务：
- 每周大概运行 agent：<次数>
- 其中多步改代码 / 排障：<比例>
- 其中文件批处理 / 数据清理 / 运维：<比例>
- 是否需要 Cloud Agents / Slack / GitHub 触发：<是/否>
- 是否需要 SSO、管理员数据控制、团队 usage metrics：<是/否>
- 是否已有 OpenAI / Anthropic / Google key 或 OpenAI-compatible gateway：<有/无>
请输出：
1. 推荐档位：Free / Build / Max / Business / Enterprise
2. 推荐账本：Warp credits / BYOK / custom endpoint / 混合
3. 升档前的 7 天验证动作
4. 哪些任务不该放进 WarpBased on my real usage in the past 30 days, recommend a Warp tier. Do not recommend by "more expensive is better."
My tasks:
- Agent runs per week: <count>
- Multi-step coding / debugging share: <share>
- File batch / data cleanup / ops share: <share>
- Need Cloud Agents / Slack / GitHub triggers: <yes/no>
- Need SSO, admin data controls, team usage metrics: <yes/no>
- Already have OpenAI / Anthropic / Google keys or an OpenAI-compatible gateway: <yes/no>
Please output:
1. Recommended tier: Free / Build / Max / Business / Enterprise
2. Recommended ledger: Warp credits / BYOK / custom endpoint / hybrid
3. A 7-day validation step before upgrading
4. Which tasks should not go into Warp
请根据我过去 30 天的真实使用，给我一个 Warp 选档建议，不要按「越贵越好」推荐。
我的任务：
- 每周大概运行 agent：约 40 次
- 其中多步改代码 / 排障：60%
- 其中文件批处理 / 数据清理 / 运维：30%
- 是否需要 Cloud Agents / Slack / GitHub 触发：否
- 是否需要 SSO、管理员数据控制、团队 usage metrics：否
- 是否已有 OpenAI / Anthropic / Google key 或 OpenAI-compatible gateway：有 Anthropic key
请输出：
1. 推荐档位：Free / Build / Max / Business / Enterprise
2. 推荐账本：Warp credits / BYOK / custom endpoint / 混合
3. 升档前的 7 天验证动作
4. 哪些任务不该放进 WarpBased on my real usage in the past 30 days, recommend a Warp tier. Do not recommend by "more expensive is better."
My tasks:
- Agent runs per week: about 40
- Multi-step coding / debugging share: 60%
- File batch / data cleanup / ops share: 30%
- Need Cloud Agents / Slack / GitHub triggers: no
- Need SSO, admin data controls, team usage metrics: no
- Already have OpenAI / Anthropic / Google keys or an OpenAI-compatible gateway: yes, an Anthropic key
Please output:
1. Recommended tier: Free / Build / Max / Business / Enterprise
2. Recommended ledger: Warp credits / BYOK / custom endpoint / hybrid
3. A 7-day validation step before upgrading
4. Which tasks should not go into Warp
Warp 官方对比 Claude Code 时，核心差别不是「谁更聪明」，而是体验边界：Claude Code 是 CLI tool；Warp 的 agent 内建在 terminal 里，有可视化 diff、Profiles、Rules、多模型与 context referencing。6注 6Note 6Warp Docs · Warp vs Claude Code —— 官方对比里，Claude Code 是 CLI tool，Warp 内置在 Warp terminal 中，提供可视化 diff、Profiles、Rules、多模型、context referencing 等集成体验。Warp Docs · Warp vs Claude Code — Warp's official comparison frames Claude Code as a CLI tool while Warp is built into the Warp terminal, adding visual diffs, Profiles, Rules, multiple models, context referencing, and an integrated experience.所以选工具时别问谁赢，问这件事在哪个表面上最自然。In Warp's official comparison with Claude Code, the key difference is not "which is smarter" but the surface: Claude Code is a CLI tool; Warp's agent is built into the terminal, with visual diffs, Profiles, Rules, multiple models, and context referencing.6注 6Note 6Warp Docs · Warp vs Claude Code —— 官方对比里，Claude Code 是 CLI tool，Warp 内置在 Warp terminal 中，提供可视化 diff、Profiles、Rules、多模型、context referencing 等集成体验。Warp Docs · Warp vs Claude Code — Warp's official comparison frames Claude Code as a CLI tool while Warp is built into the Warp terminal, adding visual diffs, Profiles, Rules, multiple models, context referencing, and an integrated experience. So do not ask which tool wins. Ask which surface this task naturally belongs on.截至 2026-07-10，这道分工题多了一个条件：「用 Warp」和「用你已付费的 CLI agent」不再互斥。Warp 会自动识别在它里面跑的第三方 CLI agent —— Claude Code、Codex、Gemini CLI、Cursor CLI、OpenCode、agy（Antigravity）等十几个 —— 并给它们叠上自己的 UI 层：rich input 编辑器、通知、code review 面板、把代码块作为上下文发进去；官方把这套能力叫 universal agent support。7注 7Note 7Warp Docs · Third-party CLI agents overview —— Warp 自动识别受支持的第三方 CLI coding agent 并叠加 rich input 编辑器、通知、code review、Remote Control 等能力，官方把这套能力叫 universal agent support；截至 2026-07-10 支持 Claude Code、Codex、OpenCode、Gemini CLI、Cursor CLI、Copilot CLI、Amp、Auggie、Droid、Pi、Goose、Antigravity（agy）、Hermes、Mistral Vibe。Warp Docs · Third-party CLI agents overview — Warp auto-detects supported third-party CLI coding agents and layers on a rich input editor, notifications, code review, Remote Control, and more; Warp calls this feature set universal agent support. As of 2026-07-10 it covers Claude Code, Codex, OpenCode, Gemini CLI, Cursor CLI, Copilot CLI, Amp, Auggie, Droid, Pi, Goose, Antigravity (agy), Hermes, and Mistral Vibe.云端一侧，Oz 也能把 Claude Code / Codex 当 harness 跑进 cloud agent（pricing 页标 beta），共享同一套 trigger、environment 与观测。8注 8Note 8Warp Docs · Harnesses in Oz —— Oz 可把 Claude Code、Codex 当云端 harness 跑，共享 Warp Agent 的 triggers、environments、secrets 与观测；推理由你提供的凭据记 provider 的账，Warp 按 compute 与 platform credits 计量；Warp pricing 页把「Use any harness in the cloud」标为 beta。Warp Docs · Harnesses in Oz — Oz can run Claude Code and Codex as cloud harnesses sharing Warp Agent's triggers, environments, secrets, and observability; inference lands on your provider account via credentials you supply while Warp meters compute and platform credits; Warp's pricing page labels 'Use any harness in the cloud' as beta.也就是说，你可以把 Claude Code 留作发动机、把 Warp 当它的驾驶舱 —— 但别把这读成「Warp 罩住一切」的定论：这是 Warp 想成为「agent 之上那层」的产品策略，你的 CLI agent 在普通终端里跑一样成立；值不值得为这层驾驶舱多养一个 GUI 终端，仍按下面这张表判断 —— 活儿在哪个表面最自然，就在哪儿做。As of 2026-07-10, the split has one new condition: "use Warp" and "use the CLI agent you already pay for" are no longer mutually exclusive. Warp auto-detects third-party CLI agents running inside it — Claude Code, Codex, Gemini CLI, Cursor CLI, OpenCode, agy (Antigravity), and a dozen or so more — and layers its own UI on them: the rich input editor, notifications, the code review pane, sending code blocks in as context; Warp calls this feature set universal agent support.7注 7Note 7Warp Docs · Third-party CLI agents overview —— Warp 自动识别受支持的第三方 CLI coding agent 并叠加 rich input 编辑器、通知、code review、Remote Control 等能力，官方把这套能力叫 universal agent support；截至 2026-07-10 支持 Claude Code、Codex、OpenCode、Gemini CLI、Cursor CLI、Copilot CLI、Amp、Auggie、Droid、Pi、Goose、Antigravity（agy）、Hermes、Mistral Vibe。Warp Docs · Third-party CLI agents overview — Warp auto-detects supported third-party CLI coding agents and layers on a rich input editor, notifications, code review, Remote Control, and more; Warp calls this feature set universal agent support. As of 2026-07-10 it covers Claude Code, Codex, OpenCode, Gemini CLI, Cursor CLI, Copilot CLI, Amp, Auggie, Droid, Pi, Goose, Antigravity (agy), Hermes, and Mistral Vibe. On the cloud side, Oz can run Claude Code / Codex as a harness inside a cloud agent (the pricing page labels this beta), sharing the same triggers, environments, and observability.8注 8Note 8Warp Docs · Harnesses in Oz —— Oz 可把 Claude Code、Codex 当云端 harness 跑，共享 Warp Agent 的 triggers、environments、secrets 与观测；推理由你提供的凭据记 provider 的账，Warp 按 compute 与 platform credits 计量；Warp pricing 页把「Use any harness in the cloud」标为 beta。Warp Docs · Harnesses in Oz — Oz can run Claude Code and Codex as cloud harnesses sharing Warp Agent's triggers, environments, secrets, and observability; inference lands on your provider account via credentials you supply while Warp meters compute and platform credits; Warp's pricing page labels 'Use any harness in the cloud' as beta. In other words, you can keep Claude Code as the engine and use Warp as its cockpit — but don't read that as a verdict that Warp wraps everything: it's Warp's play to become the layer above other agents, your CLI agent works just as well in a plain terminal, and whether that cockpit justifies keeping one more GUI terminal around is still decided by the table below — the task goes where its surface is most natural.
浏览器自动化 / 专门 Computer Use；Warp 的 cloud sandbox 只用在适配场景。Browser automation / dedicated Computer Use; Warp's cloud sandbox only when the fit is right.
Warp 的 Computer Use 是云端沙箱能力，不是接管你本机桌面。Warp's Computer Use is a cloud-sandbox capability, not control of your local desktop.
— IV
最后用五个问题做决定Make the Decision with Five Questions.
我每周是否至少有 3 件终端侧的手动活，能被 agent 接手或被 Drive 沉淀？没有，先 Free。Do I have at least 3 terminal-side manual tasks per week that an agent can take over or Drive can distill? If not, stay Free.
我撞到的是 credits 不够，还是组织控制不够？前者看 Build / Max / BYOK / custom endpoint；后者看 Business / Enterprise。Am I hitting a credits limit or an organizational-control limit? The former points to Build / Max / BYOK / custom endpoint; the latter to Business / Enterprise.
我有没有把编辑器侧长篇改写错塞进 Warp？如果是，把那部分切给 Cursor。Am I forcing editor-side long-form code changes into Warp? If yes, move that part to Cursor.
我有没有把 SSH、容器、离线环境错塞进 Warp 的产品边界？如果是，普通终端 + CLI agent 更顺。Am I forcing SSH, container, or offline work into Warp's product boundary? If yes, a plain terminal + CLI agent is smoother.
我准备上 Cloud Agent 之前，是否确认了触发、上下文、验收、额度、日志与人工回滚？没有，就先留本地。Before sending work to a Cloud Agent, have I defined trigger, context, validation, credits, logs, and human rollback? If not, keep it local.
选档看缺口，
分工看表面.
Pick the tier by the gap,
split the work by the surface.
Aklman Library
— 讨论Discussion
讨论Discussion.
评论区初始化中…Initializing comments…
01 / 01
没有匹配结果No matches.
换个关键词，或按 Esc 回到页面Try another keyword, or press Esc to return