20:"$Sreact.view_transition" 21:I[9754,["/_next/static/chunks/2t36f5p4pjzlb.js","/_next/static/chunks/1gyqeidjm5eoa.js","/_next/static/chunks/0_fjnxc4tt_l_.js","/_next/static/chunks/1ymt1shyhmu-4.js","/_next/static/chunks/3kv2zf0i7s2wc.js","/_next/static/chunks/1vs_3g_g75zmk.js","/_next/static/chunks/2g4dh802s8oz3.js","/_next/static/chunks/1bc3f0shukf1j.js","/_next/static/chunks/192kgs9v_t14j.js","/_next/static/chunks/0z6pw0v5c_lx5.js","/_next/static/chunks/08r60e6icmtkc.js","/_next/static/chunks/3trx6c0i7_x-3.js","/_next/static/chunks/40sro5bzp9bi_.js","/_next/static/chunks/1wgf3tk38y6cf.js"],"ProjectDetail"] 22:T5b4a, aidevops · AI DevOps 框架，让代理自主交付项目

aidevops / marcusquinn / MIT · Shell / 376★ · 63 fork

让代理真正把活干完，
而不是停在“好，我可以帮你”。

OpenCode 插件 · AI DevOps 框架…自动化 AI 开发运维

Vibe-Coding 很容易，DevOps 很难。aidevops 是一个 OpenCode 插件与 AI DevOps 框架，让 AI 在代码、基础设施、业务、营销、内容与创意项目中执行真正有用的工作——而不是把每个任务拖成一段冗长脆弱的对话。

GitHub ↗ 项目主页 ↗ 立即接入 ↓

监督中继 / 任务通道 pulse-supervised relay

中继地图

一个任务进来，一个成果出去Relay Map · 机制如何保证流动

STAGE 01

入口：任务进入中继

任何真实工作——一段代码、一块基础设施、一篇内容——都以一个任务的形式进入中继。它被交给一个明确负责的领域代理，而不是散落在无结构对话里。

STAGE 02

领域工作：domain work

领域代理在各自边界内推进。14 个主要代理覆盖从编码到部署的完整跨度，每一段都对齐在监督线上，而不是独自漂移。

STAGE 03

闸门检查：cross-model 校验 → gate

每一跳都经过跨模型检查：一个结果要获得另一个模型的交叉核对才被放行。闸门不开，任务不动。这是保证“下一步一定建立在真实完成之上”的纪律。

STAGE 04

收束：交付物 artifact

通过全部闸门后，任务在末端汇成一个已交付的成果。价值不在任何一个单独工具，而在于这条被持续监督、被反复校验的完整流动本身。

自主账本

不是口号，是一架真实运转的中继Autonomy Ledger · 已验证的规模

14_个主要代理 primary agents 覆盖各领域的自主执行代理

185+_条斜杠命令 slash commands 可直接调用的命令动词

2150+ 代理文件 agent files 代理 / 子代理的完整编目

1630+ 辅助脚本 helper scripts 支撑每一跳的落地能力

这些并不是装饰性的数字

它们是中继各阶段的体量：代理决定我们有多少条工作通道，文件与脚本决定每条通道能走多远，命令决定你手上有多少可拉的扳手。

14 个主要代理 = 14 条专责通道
185+ 斜杠命令 = 185+ 个可用动词
可部署到代码、基础设施、营销、内容与创意项目

监督教义

保证流动的三条闸门规则Supervision Doctrine · pulse / checks / domain

RULE 01 · PULSE

脉冲监督

贯穿整条中继的监督线持续跳动，像脉搏一样检查每个阶段是否仍在推进。没有谁脱离这条线自行其是，工作始终处在被持续监督之下。

RULE 02 · CROSS-MODEL

跨模型校验

在每一跳开闸之前，结果都要经过另一个模型的交叉核对。假设不能自定义，结论得先被校验通过，再流向下一个阶段。

RULE 03 · DOMAIN

领域代理

每个任务交给在其边界内专精的领域代理处理。专业的人（模型）就位，而不是一个通用模型硬扛所有环节。

RULE 04 · MISSION

任务工作流

从开发到部署，任务以工作流的形式贯穿整条中继。督导、闸门与分工共同把它从一个想法推进到一个交付物，而不是停在聊天。

集成路由

把旁边的系统接进这条中继Integration Routes · attachable spurs

SPUR · SERVICES

30+ 服务

从开发到部署沿途可接入的 30 多种服务，作为中继的快捷支线挂载。

SPUR · MCP

20 个 MCP 服务器

以 MCP 协议接进 20 台服务器，把外部能力变成闸门内的可用资源。

SPUR · APIS

30+ API 集成

沿中继分散挂载的 30 多条 API 支线，在各自阶段按需接入。

推荐搭配 OpenCode 与 OpenAI 模型（GPT-5.5 与 GPT-5.4 mini）运行，同时完整支持 Claude。栈取开源优先、SOTA 一切的取向，处处为速度、安全与 7×24 的结果而设计。

接入路径

现在就加入这条中继Adoption Path · install · init · update

npm install -g aidevops→ install

bun install -g aidevops→ install

aidevops init→ join

aidevops update→ sync

加入中继的三步

用 npm 或 bun 全局安装 aidevops
运行 aidevops init 初始化你的任务
用 aidevops update 保持中继最新

GitHub · 开始 ↗ 项目主页 ↗

aidevops · AI DevOps framework MIT · open-source first · SOTA everything aidevops.sh ↗

23:T567d, aidevops · An AI DevOps framework for autonomous delivery

aidevops / marcusquinn / MIT · Shell / 376★ · 63 forks

An AI DevOps framework
for autonomous delivery.

OpenCode plugin · AI DevOps framework · automate AI dev-ops

Vibe-coding is easy. DevOps is hard. aidevops is an OpenCode plugin and AI DevOps framework designed to let AI do useful work across code, infrastructure, business, marketing, content, and creative projects — without turning every job into a long, fragile chat.

GitHub ↗ Project site ↗ Join the relay ↓

Pulse-supervised relay · mission channel mission entry → delivered artifact

Relay map

A mission enters, a deliverable exitsHow the mechanism guarantees flow

STAGE 01

Entry: a mission enters the relay

Any real job — a piece of code, some infrastructure, a piece of content — enters the relay as a single mission. It is handed to a domain agent that owns it, rather than floating around in an unstructured chat.

STAGE 02

Domain work

Domain agents advance within their own boundaries. Fourteen primary agents span the full arc from coding to deployment, each aligned to the supervision line rather than drifting on its own.

STAGE 03

Check gate: cross-model validation → gate

Every hop passes a cross-model check: a result only moves on once another model has cross-checked it. No gate, no movement. It is the discipline that makes "next" always rest on something actually done.

STAGE 04

Convergence: a delivered artifact

Past every gate, the mission converges at the end into a single delivered artifact. The value is not any one tool — it is the supervised, validated flow that carries the work all the way through.

Autonomy ledger

Not a slogan — a relay that actually runsVerified scale · stage counts

14 primary agents autonomous agents across domains

2150+ agent files full catalog of agents / subagents

1630+ helper scripts the capability behind every hop

185+ slash commands verbs you can invoke directly

These numbers are not decoration

They are the size of each relay stage: agents decide how many working channels we have, files and scripts decide how far each channel can run, and commands decide how many levers sit at your fingertips.

14 primary agents = 14 dedicated channels
185+ slash commands = 185+ usable verbs
Ships to code, infra, marketing, content, creative

Supervision doctrine

Three gate rules that hold the flowPulse · cross-model · domain

RULE 01 · PULSE

Pulse supervision

A supervision line runs through the whole relay and keeps beating — a pulse that checks whether each stage is still advancing. Nothing moves off that line on its own; work always stays under continuous supervision.

RULE 02 · CROSS-MODEL

Cross-model checks

Before any hop opens, a result passes a cross-check by another model. Assumptions can't be self-declared; a conclusion must be validated before it flows to the next stage.

RULE 03 · DOMAIN

Domain agents

Each mission is handled by a domain agent that specializes in its boundary. The right model is in place for the job, instead of one general model forcing every step.

RULE 04 · MISSION

Mission workflows

From development to deployment, work runs through the relay as a mission workflow. Supervision, gates, and specialization push it from an idea to a deliverable — not to a dead-end chat.

Integration routes

Attach the systems around you to the relayAttachable spurs · services & APIs

SPUR · SERVICES

30+ services

More than thirty services you can attach along the route from dev to deploy — quick spurs on the relay.

SPUR · MCP

20 MCP servers

Wire in twenty MCP servers over the protocol, turning external capability into usable resources inside the gates.

SPUR · APIS

30+ API integrations

Thirty-plus API spurs distributed along the relay, pulled in on demand at each stage.

Recommended to run with OpenCode and OpenAI models (GPT-5.5 and GPT-5.4 mini), while Claude stays supported. The stack is open-source-first and SOTA everywhere, built for speed, security, and 24/7 results.

Adoption path

Join the relay nownpm · bun · init · update

npm install -g aidevops→ install

bun install -g aidevops→ install

aidevops init→ join

aidevops update→ sync

Three steps to on board

Install aidevops globally with npm or bun
Run aidevops init to set up your work
Keep the relay current with aidevops update

GitHub · Get started ↗ Project site ↗

aidevops · AI DevOps framework MIT · open-source first · SOTA everything aidevops.sh ↗

a:[["$","$20",null,{"enter":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-enter","default":"none"},"exit":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-exit","default":"none"},"update":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-update","default":"none"},"default":"none","children":["$","$L21","cmt2zh1ax010u62716jz0a74j:zh-CN:beautified",{"project":{"id":"cmt2zh1ax010u62716jz0a74j","slug":"marcusquinn-aidevops","name":"aidevops","description":"OpenCode 插件，自动化 AI 开发运维","authorName":"marcusquinn","authorVerification":{"verified":false,"verifiedAt":null},"category":"development_tools","subcategory":"data_ai_tools","experienceUrl":"https://www.aidevops.sh/","qualityScore":93,"featured":false,"officialSite":null,"views":{"original":{"available":true},"official":{"available":false,"reason":"disabled_by_policy"},"beautified":{"available":true,"reason":null},"defaultView":"beautified"},"beautifiedView":{"revisionId":"cmt2zh1c801166271pf2lehdn","kind":"artifact","artifact":{"schemaVersion":"localized-static-html-v4","kind":"static_html","locales":["zh-CN","en"],"defaultLocale":"zh-CN","documents":{"zh-CN":"$22","en":"$23"},"designDirections":{"schemaVersion":"dodot-design-directions-v1","directions":[{"id":"mission-relay","name":"Mission Relay","projectMetaphor":"Autonomous work as a pulse-supervised relay: a mission enters, passes through domain stages under continuous pulse supervision, cross-model checks gate each hop, and a delivered artifact exits. The framework's value is the guaranteed flow, not any single tool.","signatureVisual":"A horizontal time-line of agent stages (domain work, cross-model check, mission workflow) linked by a thin vertical 'pulse' rail that oscillates through every stage; each hop is a small gate node that must open before the mission moves on. The whole relay ends in a single converged deliverable mark.","heroGeometry":"A long horizontal conductor band across the hero: an inbound mission bead at left travels right through 4 gate nodes (each a ring with an interior cross/model check glyph), while a supervisor pulse line runs beneath the whole path and all stages align to it, signaling continuous supervision.","sectionSequence":["Relay Map (mechanism diagram: mission enters, supervised stages, delivered artifact)","Autonomy Ledger (the proven scale: 14 agents, 2150+ files, 185+ commands as relay stage counts)","Supervision Doctrine (pulse supervision, cross-model checks, domain agents explained as gate rules)","Integration Routes (30+ services, 20 MCP servers, 30+ APIs as attachable relay spurs)","Adoption Path (npm/bun install, aidevops init/update as joining a relay now)"],"surfaceModel":"Fastened plates: linear sections with visible seam lines, hairline rule borders and offset, slightly 'machined' inset panels. Motion follows the bead's travel; surfaces read as engineered track segments rather than floating cards.","palette":"Steel-to-ember: deep gunmetal blue-black field, warm ember/orange accent for the delivered artifact and gate-open states, cool slate for inactive stages, off-white text. Color encodes relay status, not decoration.","typography":"Monospace-flavored control numerals for stage counts and command verbs; a sharp grotesque for headings; restrained proportional text for body. Type rhythm follows the 1-D track rather than a grid stack.","desktopComposition":"Single dominant horizontal relay track spanning the hero; sections below alternate between a full-width mechanism strip and two-column 'ledger' panels, all tied to the same baseline so the relay line visually threads the whole page.","mobileTransformation":"The 1-D horizontal track bends into a vertical serpentine: the bead travels top-to-bottom through stacked stage cells, the pulse rail becomes a continuous left margin spine. Ledgers collapse into ticked rows. Spatial decision changes, relay concept holds.","mediaStrategy":"No external or local images exist. Media region is removed entirely; the relay strip itself is the evidence visual, driven by real counts and command verbs.","forbiddenPatterns":["No three-card metric hero row","No browser/terminal/editor shell framing","No feature-card wall","No decorative glow or gradient field as a substitute for the track","No fake product screenshot or placeholder media frame"],"compositionFingerprint":"One dominant horizontal supervised relay track whose single baseline threads every section, stage gates shown open/closed along the line, counts rendered as machine-stage numerals."},{"id":"agent-constellation","name":"Agent Constellation","projectMetaphor":"Fourteen primary agents are domains held in a supervised constellation: pulse supervision is the central gravity that keeps independent domain agents in productive orbit, and cross-model checks are the alignment that keeps their work coherent rather than drifting.","signatureVisual":"A radial constellation of domain agent nodes circling a compact supervising core; each orbit is assigned a domain, and the connecting chords between moons are the cross-model check lines that keep neighbors aligned. The system holds together through orbit, not hierarchy.","heroGeometry":"A concentric radial hero: a precise supervising core at center, three nested annular orbits carrying the 14 domain agents as small nodes; a few short chords link adjacent nodes to show checked alignment. Emphasizes a stable, symmetric, self-governing system.","sectionSequence":["Constellation Chart (the 14 domains as an orbital chart, supervisor at center)","Orbit Scales (2150+ files, 1630+ scripts, 185+ commands as layer counts per orbit band)","Alignment Rules (pulse supervision, cross-model checks, domain agents as orbital discipline)","Reach (30+ services, 20 MCP servers, 30+ APIs as the constellation's outer resource rings)","Joining Orbit (npm/bun aidevops install and update as entering the system)"],"surfaceModel":"Layered concentric surfaces: circular panels layered like star-chart planes with faint orbital guides; depth cues come from ring elevation and node size, not shadow blur. Whole page is organized in nested circular bands.","palette":"Cosmic teal-to-violet: deep navy field, teal supervisor core, violet domain moons, sparse silver guide lines. Color separates center (supervision) from satellites (domains), never used as texture.","typography":"A wide-tracked sans for the orbital labels set on circular guides; tight numbers alone for counts that sit inside orbit bands. Text follows the ring geometry rather than a marching column grid.","desktopComposition":"Hero is a large radial chart; body sections are arranged as concentric ring rows — each section a band that wraps partially around the shared center so the constellation theme persists top to bottom.","mobileTransformation":"Radial orbits flatten into nested concentric rings (bulls-eye stacks): supervisor core shrinks to a small top ring and each outer band becomes a full-width horizontal ring that the reader peels down through. Radial order maps to reading order; spatial decision changes, orbit hierarchy holds.","mediaStrategy":"No local or external media. Media region removed; the constellation chart itself is the evidence visual built from real domain and count facts.","forbiddenPatterns":["No three-card metric dashboard hero","No terminal/browser/editor shell","No flat feature-card row","No glowing particle field used only as decoration","No fabricated product image or chart"],"compositionFingerprint":"Concentric radial star-chart with a supervising core and orbiting domain nodes linked by short alignment chords, section geometry nested as rings around a persistent center."},{"id":"command-ledger","name":"Command Ledger","projectMetaphor":"The framework is a verb-led command ledger: every real command (`aidevops init`, `aidevops update`, `npm install -g aidevops`) is a row that consumes tokens and outputs a delivered outcome. The page is a working ledger of commands and their proven consequences, not a decorative terminal.","signatureVisual":"A vertical ledger of command-rows: the raw verb on the left, a token-burn column in the middle, and the proven, delivered outcome on the right; rows that pass cross-model checks are stamped as complete. Reading it shows one command 'passing' into a shipped result.","heroGeometry":"A tall right-edge ledger column in the hero: three authentic command rows aligned on a shared verb-axis, each row animated to stamp its outcome when it reaches the bottom of the column — a 'command passing to consequence' rhythm rather than a centered slogan.","sectionSequence":["Verb Ledger (real install/init/update commands as rows with outcomes)","Scale Columns (14 agents, 2150+ files, 1630+ scripts, 185+ commands as ledger totals)","How a Mission Passes (pulse supervision, cross-model checks, mission workflow as the ledger's validation pass)","Integrated Stacks (30+ services, 20 MCP servers, 30+ APIs as linked registers)","Adopt Today (run both install routes: npm and bun), then init"],"surfaceModel":"Journal-like ruled surfaces: thin ruled lines across sections, column rules demarcating verb/band/outcome, marginal annotations. Surfaces read as an auditable ledger or logbook, flat and readable, not elevated cards.","palette":"Ink-on-paper meets rail green: near-black ink on warm parchment-white, one rail-green accent for the passing/validated stamp, red reserved for 'blocked before a check'. Color only marks command state.","typography":"A tabular mono for verbs and counts so columns align at every size; a condensed grotesque for section headings. Type is the ledger itself — alignment and tabularity are the visual system.","desktopComposition":"Hero rests on a tall right ledger column; body sections are full-width ruled rows that keep the same column axes (verb | burn | outcome), so a single tabular grid governs the entire page and reads as one long auditable log.","mobileTransformation":"Column rows recompose from three side-by-side columns into stacked single-line entries (verb / burn / outcome on one line, wrapped) with the validation stamp moved to a leading checkmark — a log-line list instead of a table; identical data, different spatial reading.","mediaStrategy":"No usable local media. Media region dropped; the ledger of real commands and counts is the sole evidence visual.","forbiddenPatterns":["No terminal-shell decoration or fake prompt UI","No three-card metric hero","No feature-card wall","No random glowing grids or gradients","No invented command outputs or fake statuses"],"compositionFingerprint":"A tabular verb-ledger whose shared column axes (verb | token-burn | outcome) persist across every section, read as one continuous auditable log with passing rows stamped by validated results."}],"selectedDirectionId":"mission-relay","selectionRationale":"mission-relay best expresses the project's actual mechanism: autonomous work that travels through pulse-supervised, cross-model-checked stages to a delivered artifact. Agent-constellation is spatially striking but reads as a static network rather than the framework's movement of work; command-ledger risks the forbidden terminal reframing (a verb list can read as a decorative shell) and does not show the supervision loop. The relay places supervision and gating at the center, uses verified counts (14/2150+/185+) as stage magnitudes, and produces a composition fingerprint unlike any recent page."},"designBrief":{"schemaVersion":"dodot-design-brief-v1","selectedDirectionId":"mission-relay","projectMetaphor":"Autonomous work as a pulse-supervised relay: a mission enters, passes through domain stages under continuous pulse supervision, cross-model checks gate each hop, and a delivered artifact exits. The framework's value is the guaranteed flow, not any single tool.","signatureVisual":"A horizontal time-line of agent stages (domain work, cross-model check, mission workflow) linked by a thin vertical 'pulse' rail that oscillates through every stage; each hop is a small gate node that must open before the mission moves on. The whole relay ends in a single converged deliverable mark.","compositionFingerprint":"One dominant horizontal supervised relay track whose single baseline threads every section, stage gates shown open/closed along the line, counts rendered as machine-stage numerals.","heroGeometry":"A long horizontal conductor band across the hero: an inbound mission bead at left travels right through 4 gate nodes (each a ring with an interior cross/model check glyph), while a supervisor pulse line runs beneath the whole path and all stages align to it, signaling continuous supervision.","sectionSequence":["Relay Map (mechanism diagram: mission enters, supervised stages, delivered artifact)","Autonomy Ledger (the proven scale: 14 agents, 2150+ files, 185+ commands as relay stage counts)","Supervision Doctrine (pulse supervision, cross-model checks, domain agents explained as gate rules)","Integration Routes (30+ services, 20 MCP servers, 30+ APIs as attachable relay spurs)","Adoption Path (npm/bun install, aidevops init/update as joining a relay now)"],"surfaceModel":"Fastened plates: linear sections with visible seam lines, hairline rule borders and offset, slightly 'machined' inset panels. Motion follows the bead's travel; surfaces read as engineered track segments rather than floating cards.","palette":"Steel-to-ember: deep gunmetal blue-black field, warm ember/orange accent for the delivered artifact and gate-open states, cool slate for inactive stages, off-white text. Color encodes relay status, not decoration.","typography":"Monospace-flavored control numerals for stage counts and command verbs; a sharp grotesque for headings; restrained proportional text for body. Type rhythm follows the 1-D track rather than a grid stack.","desktopComposition":"Single dominant horizontal relay track spanning the hero; sections below alternate between a full-width mechanism strip and two-column 'ledger' panels, all tied to the same baseline so the relay line visually threads the whole page.","mobileTransformation":"The 1-D horizontal track bends into a vertical serpentine: the bead travels top-to-bottom through stacked stage cells, the pulse rail becomes a continuous left margin spine. Ledgers collapse into ticked rows. Spatial decision changes, relay concept holds.","mediaStrategy":"No external or local images exist. Media region is removed entirely; the relay strip itself is the evidence visual, driven by real counts and command verbs.","evidencePlan":["Relay Map: draw the mechanism purely from verified facts — domain work, cross-model checks, mission workflow, pulse supervision (source: summary evidence claims).","Autonomy Ledger: quote exact verified figures only: 14 primary agents, 2150+ agent files, 1630+ helper scripts, 185+ slash commands (source: summary claims).","Command rows: use only real verbs — npm install -g aidevops, bun install -g aidevops, aidevops init, aidevops update (source: readme.commands). Present as relay entry actions, not fabricated output."],"localeContract":{"zh-CN":"zh-CN 为主导文本；口号、概览、亮点与注意点取自 summary.localizations.zh-CN（如“AI DevOps 框架，让代理自主交付项目”），并与英文共享同一章节序列与构图。","en":"English mirrors zh-CN section sequence and composition with natural English copy from summary.localizations.en."},"forbiddenPatterns":["No three-card metric hero row","No browser/terminal/editor shell framing","No feature-card wall","No decorative glow or gradient field as a substitute for the relay track","No fake product screenshot, placeholder media frame, or invented command output"]},"visualReview":{"schemaVersion":"dodot-automated-design-review-v1","status":"approved","summary":"页面忠实实现了 mission-relay 设计方向：监督中继轨道作为 signatureVisual 贯穿首屏，任务珠从左端进入并依次穿过领域代理、跨模型检查、任务工作流、交付物四个闸门节点，脉冲监督线（ECG 波形）贯穿全程。章节序列（Relay Map / Autonomy Ledger / Supervision Doctrine / Integration Routes / Adoption Path）与 Brief 一致。钢灰-余烬配色、等宽数字、机械拼板表面均落实。关键事实（14 代理、2150+ 文件、1630+ 脚本、185+ 命令、30+ 服务、20 MCP、30+ API、GPT-5.5/5.4 mini、Claude）均有材料依据；安装命令全部来自 readme.commands。手机断点将横向轨道弯折为左侧连续脊线的纵向蛇形堆叠，账本折叠为单列，形成真实重构。双语结构等价且文案自然。未回退到通用 Hero/三卡/卡片墙/无依据设备壳/装饰性渐变网格，也未出现假控件。","checks":{"projectFit":{"passed":true,"evidence":"首屏实现 Supervised Relay Track：inbound mission 珠 + 四条闸门节点（domain work / cross-model / mission flow / artifact）+ 底部脉冲线 + 末端 deliverable 标记。五章节命名与 Brief sectionSequence 逐条对上（Relay Map→domains→#map，Autonomy Ledger→#ledger，Supervision Doctrine→#doctrine，Integration Routes→#routes，Adoption Path→#adopt）。铁灰蓝黑 field + ember 余烬强调色 + 石板灰未激活态 + 米白文字，符合 steel-to-ember 调色板；等宽 stage 计数与命令动词、grotesque 标题遵循 typography 约定。"},"evidenceBound":{"passed":true,"evidence":"账本区引用唯一准确数字：14 主要代理、2150+ 代理文件、1630+ 辅助脚本、185+ 斜杠命令（对应 summary 与 material facts）；集成区引用 30+ 服务、20 MCP 服务器、30+ API（summary claim）；推荐模型写为 OpenAI GPT-5.5 / GPT-5.4 mini 并支持 Claude（summary claims）。Adoption 区命令仅使用真实动词 npm install -g aidevops、bun install -g aidevops、aidevops init、aidevops update，均来自 readme.commands，未发明输出。"},"signatureVisual":{"passed":true,"evidence":"`.relay/.track` 为真实 DOM 实现：四个 .gate 环节点（含 .open 打开态、.glyph 内部 glyph）、一条 .track-line 基线、下方 .pulse SVG 心电图监督线、左端 .inbound .bead（mission）与右端 .deliver（artifact ✓）。与 Brief signatureVisual（横向时间线的 gate 节点 + 贯穿全线的 pulse 轨 + 收敛交付标记）直接对应，非占位或截图。"},"mobileTransformation":{"passed":true,"evidence":"@media(max-width:720px) 将 .track 改为纵向蛇形：.stages flex-direction:column 使阶段上下堆叠，.stage 保留 border-left 使 pulse 线成为左侧连续脊线，.track-line display:none、.deliver 转为静态行内形态。ledger/doctrine/routes/adopt 在断点内 grid-template-columns:1fr 折叠为单列，ledgrid 在 460px 收为 1 列。符合 Brief mobileTransformation（横向 track 弯成垂直蛇形、pulse 变为左缘 spine、ledger 折叠为 ticked 行）。"},"localeParity":{"passed":true,"evidence":"zh-CN 与 en 结构完全同构：各 5 个 h2、5 个 section、相同 id（map/ledger/doctrine/routes/adopt）、相同 relay 轨道结构。数字事实两语言一致（14/2150+/1630+/185+/30+/20/30+/GPT-5.5/GPT-5.4 mini）。zh-CN 为主导文本并取自 summary.localizations.zh-CN，英文为自然英文表述且语义等价，无直译生硬（如英文 'Not a slogan — a relay that actually runs' 与中文'不是口号，是一架真实运转的中继'对等）。"},"antiTemplate":{"passed":true,"evidence":"无三卡指标 Hero：首屏为单一横向轨道，非三卡行。无 browser/terminal/editor 外壳：轨道、maprow、ledger 为内容面板而非设备壳。无功能卡片墙：doctrine 与 routes 以规则/支线为主题而非通用 feature-card。渐变仅用于头发丝粗细的 .baseline 与 .track-line 线条（非底面渐变场），未以装饰渐变网格替代轨道。signatureVisual 由真实计数与命令动词驱动，media region 被移除。"},"noFakeUi":{"passed":true,"evidence":"QA 报告两语言两视口 fakeControlCount 均为 0。互动元素均为真实链接（GitHub https://github.com/marcusquinn/aidevops、项目主页 https://www.aidevops.sh/、页内 #adopt 锚点与 .cta）。relay 轨道 aria-hidden=true 属装饰；.gate 为含文本 glyph 的 div 非交互控件；Adoption 行尾 '→ install/join/sync' 为语义状态标注（描述命令作用），非仿冒终端输出或假按钮。"}},"issues":[]},"repairEvidence":[],"provenance":{"executor":"opencode","model":"deepseek/deepseek-v4-flash","opencodeVersion":"1.1.58","promptVersion":"dodot-showcase-deepseek-v4","inputHash":"4cf3d37b4427cec67df071d74b54749c3ef327fa11c1a0a3cb2c33bdffa3013d","artifactHash":"d7ed2e0c072428aa3c1fcc547506f25af547260eeaf8fa259b634824bc0ac86b","designSkill":"dodot-showcase-design"},"qaEvidence":{"zh-CN":[{"viewport":"desktop","passed":true,"localePassed":true,"metrics":{"language":"zh-CN","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":2243,"cjkCount":1010,"latinLetterCount":709,"scrollWidth":1280,"clientWidth":1280,"scrollHeight":3763,"h1Top":151,"bottomInset":219.234375,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"feca3926be2d5364ca3e50e56c4e7cba65240f8966d20750b9a4ed6b22250044","middle":"02cbbd479c5b739de4f8b39e23d70948fa99f0665a6bb32acca03503e4c8c448","bottom":"864535e2583a94174925bfaf06ed18b4a6672b35bc11c8932744ac509369dcf7"},"consoleErrors":[],"networkRequests":[]},{"viewport":"mobile","passed":true,"localePassed":true,"metrics":{"language":"zh-CN","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":2243,"cjkCount":1010,"latinLetterCount":709,"scrollWidth":390,"clientWidth":390,"scrollHeight":5978,"h1Top":178,"bottomInset":171.78125,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"a63a03b57c0a8861aaf332820cb90d3b9627a57fc3bdaf0980ed6bbc345408f2","middle":"280319eafbebfcf014e046db8e81efbe4c1ca71e3fc6269ea4c5bd051b222f53","bottom":"a190a2936105bddde6aca38212f807d768d99eb54856ec628c24fc4ea95947b4"},"consoleErrors":[],"networkRequests":[]}],"en":[{"viewport":"desktop","passed":true,"localePassed":true,"metrics":{"language":"en","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":4514,"cjkCount":0,"latinLetterCount":3518,"scrollWidth":1280,"clientWidth":1280,"scrollHeight":4079,"h1Top":151,"bottomInset":219.296875,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"d1dceebff35ffc9980bcc29589b54a32f01f9fed92e54a43a9fe37f5de5ea4fa","middle":"ae13ffdb40cca756ff7052806dd7a428abfaa4f0a0b22c704a24e0e500bedc9f","bottom":"93ff7c1246c1d18058dc9c61097f9aebcc096afdd60b7742afd973b057d58c6e"},"consoleErrors":[],"networkRequests":[]},{"viewport":"mobile","passed":true,"localePassed":true,"metrics":{"language":"en","h1Count":1,"h2Count":5,"sectionCount":5,"textLength":4514,"cjkCount":0,"latinLetterCount":3518,"scrollWidth":390,"clientWidth":390,"scrollHeight":6867,"h1Top":178,"bottomInset":171.609375,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"ec87852b1cf66b8a5fbbafa4b357787b4fbcc796b1b263932da86e161b3bfbb3","middle":"37cfbc57f2df98815245d187ca573928ad8c8825eb22382daf2993045e957895","bottom":"59cfdf0eecfac3bd71aa6024ded24e432b31faa1a0c600735b635b65d0a9a3b6"},"consoleErrors":[],"networkRequests":[]}]}},"generatorVersion":"opencode-1.1.58:deepseek/deepseek-v4-flash:dodot-showcase-deepseek-v4","publishedAt":"2026-08-20T07:03:25.793Z"},"rawExtraction":{"projectRevisionId":"cmt2zh1bz01106271gdycidml","projectVersion":1,"sources":[{"canonicalKey":"github:marcusquinn/aidevops","role":"primary","revisionVersion":1,"url":"https://github.com/marcusquinn/aidevops"}],"facts":[{"key":"repository_owner","value":"marcusquinn","evidence":[{"path":"repository.owner","excerpt":"marcusquinn"}]},{"key":"repository_name","value":"aidevops","evidence":[{"path":"repository.name","excerpt":"aidevops"}]},{"key":"description","value":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic.","evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]},{"key":"primary_language","value":"Shell","evidence":[{"path":"repository.language","excerpt":"Shell"}]},{"key":"license","value":"MIT","evidence":[{"path":"repository.license","excerpt":"MIT"}]},{"key":"topics","value":"agents, agents-md, ai, ai-agents, ai-assistant, ai-assistants, ai-assisted-development, ai-code-review, ai-coding, ai-dev-ops, ai-devops, ai-system-admin, claude, devops, devops-tools, git-workflow, git-worktrees, opencode, opencode-plugin","evidence":[{"path":"repository.topics","excerpt":"agents, agents-md, ai, ai-agents, ai-assistant, ai-assistants, ai-assisted-development, ai-code-review, ai-coding, ai-dev-ops, ai-devops, ai-system-admin, claude, devops, devops-tools, git-workflow, git-worktrees, opencode, opencode-plugin"}]},{"key":"source_count","value":"1","evidence":[{"path":"projectRevision.sources","excerpt":"github:marcusquinn/aidevops"}]}],"taxonomy":{"category":"development_tools","subcategory":"data_ai_tools","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]},"lifecycle":{"stage":"prototype","confidence":0.65,"evidence":[{"path":"readmeExcerpt","excerpt":" ![AI DevOps social graph preview](docs/assets/og-image.png) # AI DevOps Framework **[aidevops.sh](https://aidevops.sh)** is an [OpenCode](https://opencode.ai/) plugin and AI DevOps framework for people who want AI to do useful work across code, infrastructure, business, marketing, content, and creative projects without turning every job into another long, fragile chat. Most AI tools still leave you doin"}]},"userIntents":[{"slug":"build_software","label":"构建或改进软件","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]},{"slug":"automate_work","label":"自动化工作流程","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]},{"slug":"create_content","label":"创作内容或设计","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]},{"slug":"analyze_data","label":"分析与理解数据","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]},{"slug":"learn_explore","label":"学习与探索","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"Vibe-Coding is easy. DevOps is hard. OpenCode & Git token-efficient AI agent automation for your app, business, and personal development. Opinionated tools, services, CLI & API stack for speed, security, and 24/7 results. Open-source first. SOTA everything. Try on your repos for money-making magic."}]}],"readme":{"headings":["AI DevOps Framework","**The Aim**","**Why This Framework?**","Attribution","Quick Reference","Key Commands","Agent Structure","What You Can Ask aidevops To Do","Recent framework upgrades","**Enterprise-Grade Quality & Security**","Report creation, previews, and PDF exports","**Security Notice**","Security Commands","**Quick Start**","Installation Options","Optional Design Artifact Studio","Inspect optional companion status","Print safe install plan only","Install alongside aidevops only after opting in","Start through aidevops local HTTPS if Open Design only prints localhost"],"commands":["npm install -g aidevops && aidevops update","bun install -g aidevops && aidevops update","npm install -g markdownlint-language-server # Markdown diagnostics","cargo install toon-lsp # TOON syntax validation","npx repomix","npx repomix --compress","npm install -g @augmentcode/auggie@prerelease","pip install llm-tldr","npm install -g aidevops && aidevops update","bun install -g aidevops && aidevops update"]},"changeSet":[{"field":"repository","beforeHash":"871aacc6dcde84fb4197ef6fab84a7bc1e92264e3106f4a1db6877cd94f69a81","afterHash":"c1b355a76e9aab924a55116fbcc0ceb2b929681e418e788a6a2994665e8ab062"}]},"publishedAt":"2026-08-16T12:08:30.817Z","updatedAt":"2026-08-21T13:26:51.665Z","summary":{"id":"cmt2zh1c201126271gdjrmkxo","projectType":"product","oneLine":"OpenCode 插件，自动化 AI 开发运维","overview":"aidevops 是一个 OpenCode 插件和 AI DevOps 框架，旨在让 AI 在代码、基础设施、业务、营销、内容和创意项目中执行有用的工作，而无需将每个任务变成冗长脆弱的对话。它提供 14 个主要代理、2150+ 代理文件、1630+ 辅助脚本和 185+ 斜杠命令，支持 30+ 服务、20 个 MCP 服务器和 30+ API 集成。功能包括脉冲监督、领域代理、跨模型检查和任务工作流，帮助用户管理从开发到部署的完整项目生","highlights":["14 个主要代理","2150+ 代理文件","185+ 斜杠命令","30+ 服务集成"],"bestFor":["希望自动化 AI 开发运维的开发者","需要管理复杂 AI 代理工作流的团队","寻求 OpenCode 增强插件的用户"],"caveats":["推荐使用 OpenAI 模型，但 Claude 也支持","需要熟悉 OpenCode 和 CLI 环境","功能丰富，学习曲线可能较陡"],"categorySuggestion":"development_tools","subcategorySuggestion":"coding_agents","lifecycleStages":["idea_requirements","product_design","frontend","backend_data"],"audienceIntents":["build","integrate","operate","evaluate"],"confidence":0.85,"evidence":[{"claim":"aidevops 是一个 OpenCode 插件和 AI DevOps 框架，用于在代码、基础设施、业务、营销、内容和创意项目中自动化 AI 工作。","sourceType":"readme","locator":"https://github.com/marcusquinn/aidevops"},{"claim":"它提供 14 个主要代理、2150+ 代理/子代理文件、1630+ 辅助脚本和 185+ 斜杠命令。","sourceType":"readme","locator":"https://github.com/marcusquinn/aidevops"},{"claim":"支持 30+ 服务、20 个 MCP 服务器和 30+ API 集成。","sourceType":"readme","locator":"https://github.com/marcusquinn/aidevops"},{"claim":"包含脉冲监督、领域代理、跨模型检查和任务工作流等功能。","sourceType":"readme","locator":"https://github.com/marcusquinn/aidevops"},{"claim":"推荐使用 OpenCode 和 OpenAI 模型（GPT-5.5 和 GPT-5.4 mini），同时支持 Claude 模型。","sourceType":"readme","locator":"https://github.com/marcusquinn/aidevops"},{"claim":"安装方式为 `npm install -g aidevops`，并提供 CLI 命令如 `aidevops init` 和 `aidevops update`。","sourceType":"readme","locator":"https://github.com/marcusquinn/aidevops"},{"claim":"最新版本为 v3.32.269，发布于 2026-08-17。","sourceType":"release","locator":"https://github.com/marcusquinn/aidevops/releases/tag/v3.32.269"}],"publishedAt":"2026-08-17T13:32:14.287Z"},"github":{"url":"https://github.com/marcusquinn/aidevops","ownerAvatarUrl":"https://avatars.githubusercontent.com/u/6428977?v=4","starsCount":376,"forksCount":63,"language":"Shell","syncedAt":"2026-08-17T13:29:03.515Z","license":"MIT","defaultBranch":"main","archived":false,"disabled":false,"lastChangedAt":"2026-08-17T13:20:49.000Z","missing":false},"tags":[],"updates":[]},"entryPoint":"discovery","claim":{"state":"unclaimed","verified":false},"initialView":"beautified","community":{"viewer":null,"state":{"likesCount":0,"creativeCount":0,"followersCount":0,"commentsCount":0,"liked":false,"creative":false,"followed":false},"comments":{"comments":[],"nextCursor":null}}}]}],"$L24"]