Agents
elevenlabs/skills
Build voice AI agents with ElevenLabs. An agent skill from elevenlabs/skills.
arkcli 模型查询与基础模型服务激活能力:列出、搜索、获取火山公共基础模型详情,以及用户明确要求的开通/激活模型服务(arkcli models activate);Volc 还支持 TTFT/TPOT 性能排名、延迟趋势和输入长度对比。激活已有基础模型服务不等于部署/创建 Endpoint,不得转成 +deploy。优先使用产品命令 arkcli models ...,而不是直接调用…
$ npx skills add volcengine/ark-cli --skill arkcli-models -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install volcengine/ark-cli arkcli-models --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/volcengine/ark-cli.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/arkcli-models .claude/skills/arkcli-models && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "arkcli-models" agent skill from https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-models into .claude/skills/arkcli-models/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "arkcli-models", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-modelsType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add volcengine/ark-cli --skill arkcli-models -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install volcengine/ark-cli arkcli-models --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/volcengine/ark-cli.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/arkcli-models .agents/skills/arkcli-models && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "arkcli-models" agent skill from https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-models into .agents/skills/arkcli-models/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "arkcli-models", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add volcengine/ark-cli --skill arkcli-models -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install volcengine/ark-cli arkcli-models --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/volcengine/ark-cli.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/arkcli-models .cursor/skills/arkcli-models && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "arkcli-models" agent skill from https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-models into .cursor/skills/arkcli-models/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "arkcli-models", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/volcengine/ark-cli.git --path skills/arkcli-models--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add volcengine/ark-cli --skill arkcli-models -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install volcengine/ark-cli arkcli-models --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/volcengine/ark-cli.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/arkcli-models .gemini/skills/arkcli-models && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "arkcli-models" agent skill from https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-models into .gemini/skills/arkcli-models/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "arkcli-models", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install volcengine/ark-cli arkcli-modelsInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add volcengine/ark-cli --skill arkcli-models -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/volcengine/ark-cli.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/arkcli-models .github/skills/arkcli-models && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "arkcli-models" agent skill from https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-models into .github/skills/arkcli-models/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "arkcli-models", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add volcengine/ark-cli --skill arkcli-models -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install volcengine/ark-cli arkcli-models --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/volcengine/ark-cli.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/arkcli-models .opencode/skills/arkcli-models && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "arkcli-models" agent skill from https://github.com/volcengine/ark-cli/tree/main/skills/arkcli-models into .opencode/skills/arkcli-models/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "arkcli-models", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
arkcli-modelsarkcli 模型查询与基础模型服务激活能力:列出、搜索、获取火山公共基础模型详情,以及用户明确要求的开通/激活模型服务(arkcli models activate);Volc 还支持 TTFT/TPOT 性能排名、延迟趋势和输入长度对比。激活已有基础模型服务不等于部署/创建 Endpoint,不得转成 +deploy。优先使用产品命令 arkcli models ...,而不是直接调用…
Arkcli Models is an agent skill from volcengine/ark-cli. arkcli 模型查询与基础模型服务激活能力:列出、搜索、获取火山公共基础模型详情,以及用户明确要求的开通/激活模型服务(arkcli models activate);Volc 还支持 TTFT/TPOT 性能排名、延迟趋势和输入长度对比。激活已有基础模型服务不等于部署/创建 Endpoint,不得转成 +deploy。优先使用产品命令 arkcli models ...,而不是直接调用 Raw API。反触发:用户的最终目标是创建 / 部署 Endpoint 时,本 skill 承担有界的只读候选查询,owning skill 由创建路径确定:普通产品创建走 arkcli-deploy;用户显式要求 raw CRUD / CI / 无守卫的 infer endpoint create 走 arkcli-infer-endpoint。候选首先执行 models search ... --size 10 --format json;成功完整结果直接复用,空结果与捕获缺损按 reference 有界恢复,并把实时返回的 name 与非空 primaryversion 组合成可直接传给 --model 的完整…
Its SKILL.md is about 4.5k tokens, which your agent loads only when the skill is triggered. The skill folder holds 8 other files, including reference files (for example `references/arkcli-models-activate.md`, `references/arkcli-models-get.md` and `references/arkcli-models-list.md`).
It sits in Backend & APIs, covering REST APIs. The repository describes itself as: The fastest way to put Volcengine Ark in your terminal and your AI agent — go from prompt to generated media, multimodal answer, or deployed endpoint in a single command, no API… The licence is Apache-2.0.
4 steps, taken from the step headings in SKILL.md.
Read from SKILL.md and the folder at commit fb5b7be. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
No scripts in the folder and no shell commands in SKILL.md (its code samples are bash).
From the folder's file list and the shell code blocks in SKILL.md.
Links to these hosts (documentation or services it may open):
ark.volcengine.comFrom URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Arkcli Models loads about 4.5k tokens when it runs, and up to ~18k if it reads all its reference files. Until then it costs about 160 tokens; SKILL.md has 1,207 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
The full file from volcengine/ark-cli at commit fb5b7be, republished under its Apache-2.0 licence (© volcengine). 1,207 words, ~4,457 tokens.
.claude/skills/arkcli-models/SKILL.md (or your agent's skills folder). This skill also uses 7 other files; get the full folder from GitHub.CRITICAL — 开始前 MUST 先用 Read 工具读取 ../arkcli-shared/SKILL.md,其中包含认证闸门、配置排查与命令选择顺序
CRITICAL — 若模型查询服务于 Endpoint 创建,必须先完成 owning Skill 交接:普通创建先用 Read 读取 ../arkcli-deploy/SKILL.md;显式 raw CRUD / CI / 无守卫创建先用 Read 读取 ../arkcli-infer-endpoint/SKILL.md。未读取 owning Skill 前,禁止执行认证检查、模型查询或任何其他命令。
CRITICAL — 所有 models 命令在执行之前,务必先用 Read 工具读取其对应的 reference 文档,禁止直接盲目调用命令。
models search 就把模型查询当成最终任务。arkcli-deploy;显式 raw CRUD / CI / 无守卫创建读取并保留 arkcli-infer-endpoint。arkcli models ...shortcuts/models/../arkcli-api-explorer/SKILL.md+chat / +gen / Endpoint 创建挑模型,先加载并保留对应上游 skill,本 skill 只提供只读查询。普通 Endpoint 创建的 owning skill 是 arkcli-deploy;显式 raw CRUD / CI / 无守卫的 Endpoint 创建意图必须切换到 arkcli-infer-endpoint,并在查询候选前先加载它pricing models --modality Audio,仅按当前计价项作答;未返回就说明价格源未覆盖。models search 再用目录里的 Published/Retiring 反驳公告。若用户未说明使用 Agent Plan、Coding Plan 还是按量 Endpoint,分别标注各产品适用范围;Agent Plan 先看模型下线公告,Coding Plan 看模型下线公告,按量 Endpoint 看平台模型下线公告。只引用该产品公告中旧型号的具体行、时区与迁移目标,再检查较新公告行中建议迁移型号是否也在下线,给出仍受支持的下一站;不能把某套餐的日期套到另一产品。不明产品时给条件化结论。arkcli-plans / arkcli-infer-endpoint owning Skill;公告回答后,才按用户实际配置区分 auto/ark-code-latest、写死的 Model Name、自建 ep-...。账号实时状态查询失败则明确未知,不能以公共目录状态代替。纯问公共目录生命周期才继续本 skill 的 models search 流程。arkcli-custommodel,套餐模型清单走 arkcli-plans+chat / +gen / +deploy 需要先确定可用模型名arkcli-deploy;只有用户显式指定 raw CRUD / CI / 无守卫创建时转 arkcli-infer-endpointarkcli-api-explorerarkcli-doctor;公共榜单不能用于 Endpoint 诊断arkcli-auth 或 arkcli-configarkcli models 当默认入口;非模型问题不要先查模型+chat / +gen 找正确模型+deploy 确认可部署模型先读 references/arkcli-models-get.md,使用 arkcli models get <model-id> --format json,读取 pricing.prices。按 service_type、label、dimensions、usage_unit、unit_code 和可选 usage_period 判断适用价格,不取第一条,不再读取旧 pricing.charge_items / pricing.multi_charge_items。price: null 表示缺价,0 表示该条件下零价,空数组不代表免费或未开通;开通和权益仍看 pricing.state、pricing.inference_free_usage 等独立字段。--version 选择详情版本,价格按返回的基础模型 name 查询,不保证为版本专属报价。
用户已给出完整模型 ID,并要求核对 Responses API 等某个 API capability 时,这是「单模型事实核对」,不是候选搜索:
arkcli models get <model-id> --format json,读取完整 api_support。models search 代替;search 是候选发现和重排视图,不是精确模型 API 支持度的最终事实源。api_support 数组中按 name / key / path 定位目标 API,只按该项 supported 下结论;项缺失时报告元数据不足。arkcli-chat)解释「模型声明支持」与「当前 Endpoint / 账号可访问」是两层事实;不用真实调用反复试错。这是任何"按意图找模型"需求的第一动作,先于 search / list / resources 的命令选择。MUST 先用 Read 打开 references/arkcli-models-scenario-table.md 判定意图是否命中某个场景标签,然后才决定跑什么命令。 这张人工策展的 场景标签 → 推荐模型 表是意图排序的最高权重信号,压过下方一切命名启发式 / modality / capability 过滤。
易被截胡的反例(必读)——下面这些"闻起来像硬指标"的意图,本质是场景标签,必须先查场景表,不要本能地跳到 --capability / --modality 降级线:
| 用户这么说 | ❌ 别直接 | ✅ 第 0 步命中的场景标签 → 推荐 |
|---|---|---|
| "复杂推理 / 多步骤 / 效果最强的模型" | search --capability thinking | 复杂推理 / Agent 任务 → 查表,再以实时 lifecycle 排除下线候选 |
| "做图片生成用哪个模型" | resources list --modality image 就收手 | 图片生成 → doubao-seedream-5-0 |
| "做视频 / 角色扮演 / 字段抽取" | search --modality ... | 视频生成 / 角色扮演 / 信息抽取 → 查表 |
命中后按表里 JOIN 协议执行:取推荐模型族名词干 → arkcli models search <族名词干> 回左表校验事实 → 置顶推荐 + 补 2–3 备选。表里的版本/full-id 只是起点,命中以 search 实时返回为准(实测表 id 会陈旧)。
只有这两种情况才跳过场景表、降级到下方 search + 命名启发式:
注意区分:"复杂推理"是场景标签(先查表),不等于用户给了
thinking硬指标。前者走第 0 步并校验实时 lifecycle,后者才走降级线;下线中的表内旧型号不能成为新接入推荐。
层级边界(重要,别串台):场景表是"模型广场选型层",回答"这个意图在广场上选哪个模型",数据源永远是 models search(catalog)。 它给的是 catalog 推荐模型名(如生图 → doubao-seedream-5-0)。
resources list → models get → +gen / +chat 负责(arkcli-gen 已实装),不在本表职责内。models search,不要跑去查 resources list 把"选型"做成"查可用资源"。只有用户真的要生成 / 调用时,才下沉到 resources list 解析本 profile 的真实 id。这是意图排序线专属。枚举 / 盘点 / 统计是另一条线,走 list,不挂场景表重排。
search —— 它做关键词模糊 + modality/context/capability 结构化过滤listmodels list --page-all;“我上传/精调的模型”、cm-* → arkcli custommodel list/get;“我的 Plan 支持哪些模型” → plans model-list --plan <plan>。“我的模型”范围不明时先澄清,不拿公共目录冒充账号资产。references/arkcli-models-performance.md,再使用 models performance rank、trend 或 input-lengthget+chat / +gen / Endpoint 创建找模型:先保留 owning skill,初始只执行一次有界 models search <keyword> --size 10 --format json(无关键词时省略关键词),成功完整结果形成候选后立即回原任务;不逐候选循环 get。空结果、缺损捕获或需核对精确 API 的情况按 search 恢复规则 处理;不得把一次查询预算变成遗漏关键证据的理由。activate,先读 references/arkcli-models-activate.md;如果用户只是要 deploy / 创建端点,由 deploy / infer-create 自行触发隐式开通即可,不要先单独 activatearkcli auth statusarkcli models search + 对应 flag(--modality、--min-context-window、--capability、--cache-type...);缓存能力必须看 cache_types,不要使用旧 capabilities.cachingarkcli models search <keyword>(默认返回全部命中,无分页)arkcli models listarkcli-custommodel 查询账号资产;公共目录日期统计才使用 models list --page-all。过滤字段以所选命令实际输出为准,不能混用两类 schemaarkcli models getarkcli models activate <name> [--sub-services ...];CI 场景加 --yes,本地请求预览用 --dry-run,但不得称为服务端校验+chat / +gen / +deploy先保留用户给出的精确 Endpoint / Custom Model ID,以及当前资源目录返回的合法 Plan 调用名。只有需要公共基础模型的版本化 ID 时,才从本轮 name 与非空 primary_version 组合;Plan 的 output_name / 路由别名不是“漏版本”,不得机械追加日期。实际调用遵守 owning Skill 的执行上下文,目录命中本身不保证账号可调用。
primary_version 格式不固定,不要用正则自行判断"看起来像完整 ID"。实际观察到的格式分布(~152 个模型中约半数非 6 位):
| 格式 | 样例 full ID | 常见家族 |
|---|---|---|
YYMMDD(6 位日期) | doubao-seed-1-6-251015 | doubao 系 |
YYYYMMDD(8 位日期) | qwen3-14b-20250429、glm-4-6-20250930 | qwen / glm 开源 |
| 限定前缀 + 日期 | doubao-seed-2-0-code-preview-260215、doubao-seed-1-6-nano-unfiltered-250928 | 灰度 / 变体 |
| 纯路由字符串 | kimi-k2-6-modelhub、open-source-models-default | 外部接入 |
| 短数字 | qwen3-235b-a22b-instruct-2507 | qwen 部分 |
| 空串 | doubao-seed-tts-2-0(full ID 就等于族名) | TTS / ASR 部分 |
两条获取路径,按效率递减:
arkcli models search <keyword> 或 arkcli models list 时,返回 JSON 每个 item 自带 primary_version 字段,Agent 直接读取并拼接 <name>-<primary_version>,无需额外调用。search 还会同步返回 context_window / input_modalities / output_modalities / capabilities / cache_types 等结构化字段(来自 ArkModels enrich),下游可以直接基于此判断模型是否合适,省去再调 get 的成本。arkcli models get <name> --transform 'primary_version' 直接返回版本号注意:--transform 输出带 JSON 双引号(如 "251228"),shell 拼接前必须 tr -d '"' 剥掉:
VER=$(arkcli models get doubao-seed-1-8 --transform 'primary_version' | tr -d '"')
# VER=251228 -> full ID: doubao-seed-1-8-251228
# VER=""(空串模型)-> 直接用族名: doubao-seed-tts-2-0
if [ -n "$VER" ]; then
MODEL="doubao-seed-1-8-$VER"
else
MODEL="doubao-seed-1-8"
fi
arkcli +chat --model "$MODEL" "你好"models search [--modality / --capability ...] -> +chat / +gen
models search [--min-context-window / --capability ...] -> +deploy
# 找最新最强的 200K+ 文本模型,含 thinking 能力
arkcli models search --modality text --min-context-window 200000 --capability thinking --strict-filter
# 找视频生成模型
arkcli models search --modality video --strict-filter
# 找多模态 LLM
arkcli models search --multimodal --output-modality text --strict-filter--strict-filter 强烈建议带上:默认行为是"缺数据保留"(避免误杀),加 strict 后只返回 100% 满足条件的模型。
用户问参数支持、缺失参数的运行时行为、search/get 冲突或 NotFound 时,先读 get 的证据与排障规则。缺字段不能当否定证据,冲突必须分别标明来源;公共目录不能用来盘点账号自定义资产。
models search/get 证明公共目录与元数据,不证明 Endpoint 已创建、控制台候选可见、Key 有效、余额充足或当前账号有权限。绑定问题先用 infer endpoint get / resources resolve 获取真实绑定,再查模型详情;Custom Model 的 base_model_* 是 lineage,不能替换实际 cm-* 绑定。api_support 按目标 API 项的 supported 判断;项缺失、capability 为 null/unknown 都是“证据不足”,不是 true 或 false。图像输入能力看 input_modalities,文本输出不能反证不支持读图。目录能力与本次调用结果分开报告。context_window 保留精确整数用于筛选。展示缩写按 1K = 1024、1M = 1048576 tokens:131072 → 128K,262144 → 256K,1048576 → 1M,同时给原值;不能整除时直接给精确 token 数,不把 262144 写成 262K。get reference 解析,不用“尾部六位数字”自行拆 ID,不靠换家族或反复激活修复 NotFound。../arkcli-auth/SKILL.md../arkcli-config/SKILL.md+chat、+gen、+deploy 找模型名,应优先完成"查模型后回到原任务",不要停留在 models 本身| 命令 | 说明 |
|---|---|
arkcli models search [keyword] [filters] | Agent 首选:全量召回 + ArkModels enrich + modality/context/capability/cache type 结构化过滤 + 重排;返回字段含 context_window / input_modalities / output_modalities / capabilities / cache_types |
arkcli models list | 按 modality 全量枚举、翻页统计、模型资产盘点;轻量,不含 enrich |
arkcli models get <id> [version] | 单个模型完整详情(聚合多 API,最重也最全) |
arkcli models performance {rank,trend,input-length} | 仅 Volc:公共模型性能排名、趋势与输入长度对比 |
优先级口诀:场景表命中(★★★) > 命名启发式(★★,本节) > update_time。 先走「快速决策 第 0 步」的场景表;只有场景表没覆盖该意图、或用户点名第三方/开源/历史模型时,才用本节的命名启发式排序。
search 加 filter 之后通常还会剩多个候选,命名里有规律的 tier 信号可以帮 agent 做最后一步排序。这是启发式,不是硬规则:硬指标(context_window / capabilities / modality)永远优先于命名。
2-x > 1-8 > 1-6 > 1-5 > ...代次跳跃通常强于 tier。例:doubao-seed-2-0-mini 在多数任务上强于 doubao-seed-1-6-pro,因为基座模型代次差距大于尺寸档差距。
pro ≥ 无后缀 > lite > flash > mini > nanopro:旗舰,最大尺寸 / 最强能力doubao-seed-1-8):主力档,通常 ≈ prolite:成本/性能均衡flash:速度优化mini / nano:低延迟、高并发、最低成本-code → 编程优化(Doubao-Seed-Code 等)
-thinking → 思考能力强化(更长 reasoning)
-vision → 视觉理解强化
-character → 角色扮演 / 长旁白
-translation → 翻译专用
-tts / -voice → 语音合成
-asr → 语音识别specialty 模型在它的领域内通常强于同代通用 pro,但跨任务时不可移植。
primary_version 比较的注意事项260215 > 251228 > 251015)qwen3-14b-20250429(8 位)和 doubao-seed-1-6-251015(6 位)数值上前者大,但和"哪个强"无关update_time 字段(ISO 8601,可字典序比较),不要去 parse primary_versionfilter (硬指标,必须满足)
↓
代次 (2-x > 1-x,跨代差距通常 > 同代 tier 差距)
↓
tier (pro > 无后缀 > lite > flash > mini > nano)
↓
update_time (同代同 tier 时,新者优先)search 命令里实装不用 agent 再做客户端 reorder —— arkcli models search 按"先 family 分组、后组内细排"的两阶段排序:
1. bucket — keyword 可见性 (name 命中 > desc 命中 > 兜底 > hidden)
↓
2. 跨 family: 按 family 的"代表分"排(family 成员中的 max ctx → max update_time → family 名)
↓ → 同 family 的所有成员保持在一起
3. 同 family 内部: gen DESC → tier DESC → ctx DESC → update_time DESC → name ASC关键设计:
context_window=null),也会浮在已有 ctx 数据的旧版本之前。例:glm-5-1(gen=501, ctx=null)排在 glm-4-6(gen=406, ctx=262144)之前;kimi-k2-6(gen=600, ctx=null)排在 kimi-k2-5(gen=500, ctx=262144)之前。doubao-seed-2-0-mini(gen=200, tier=50)排在 doubao-seed-1-8(gen=108, tier=80)之前 —— 跨代差距通常大于同代 tier 差距。被打了 体验隐藏 / 推理隐藏 / 广场隐藏 任一标签的旧模型自动沉底(不会被 list/search 屏蔽,但永远在结果末尾)。
说明:
体验隐藏/推理隐藏/广场隐藏是火山方舟模型平台customized_tags中的旧版页面隐藏标签,与本 CLI 命令无关,arkcli 仅读取它们用于排序。
ArkModels 给公共模型目录记录打 lifecycle_status,三种值,Search 处理方式不同。这只描述目录视图,不是 Agent/Coding Plan 套餐或个人 Endpoint 的停服事实源;目录 Published 不足以推翻产品下线公告:
| status | 含义 | search 默认行为 |
|---|---|---|
Published | 公共目录正常展示,不代表任一套餐仍支持 | ✓ 显示 |
Retiring | 公共目录正在下线,不建议新接入;具体可调用性仍看产品与账号 | ✓ 显示,agent 应口头提醒并找替代 |
Shutdown | 公共目录已下线;不能据此推断另一产品的独立时间表 | ❌ 默认过滤掉(加 --include-deprecated 才回来) |
另外,display_name 含 废弃 / 下线 / 已下架 / deprecated 关键词的模型也按 Shutdown 处理(兜底,因为有些遗留模型不在 ArkModels 元数据里,靠人工标记)。
Agent 行为约定:
lifecycle_status="Retiring" 的模型时,主动提醒用户它正在下线,并尝试在同 family 里找一个更新版本(用 search + 正确 keyword 即可)下面这些行为是错的 —— 命令选错会让 agent 拿不到 enrich 数据(context_window / modalities / capabilities)、得不到 hidden 沉底 / 加权重排,进而给出错误推荐。
list 找模型 —— 找模型(任何"哪个模型/找一个 X 模型"意图)一律走 search。list 不做关键词模糊、不做加权重排、不返回 enrich 字段。list --modality video 选生视频模型 —— 用 arkcli models search --modality video --strict-filter,它结合 ArkModels 的 output_modalities 和 task_types 兜底,召回更准且能进一步组合 --min-context-window / --capability 过滤。list --modality text 选 LLM —— 同上,用 search --modality text --strict-filter。list --modality text 只看 foundation_model_tag.filter_domains 一层信号,会漏掉很多模型。list 再 get 来验证 context_window 等参数 —— search 已经在结果里直接返回 context_window / max_input_tokens / max_completion_tokens,省掉一次 get 调用。search 不传 keyword —— 现在不传 keyword 是返回全量 152 条按 UpdateTime 降序,不再是策展热门。需要少量结果用 --size 5。lifecycle_status="Retiring" 的模型推荐给用户做新接入 —— 这些虽然还能调,但 vendor 已经标记下线倒计时。看到 Retiring 候选时主动提示并搜更新版本。search 上做客户端二次过滤来弥补 list 的不足 —— 直接用 search 自带的 --modality / --min-context-window / --capability flag。get。模糊名称或找候选才先 search。models list 没有服务端时间过滤 flag 就去 arkcli api --list 探 Raw API —— 先 models list --page-all --format json,再用本地 JSON 处理按 create_time 过滤。arkcli models list --help 当前版本确实没有可枚举输出,且本地 JSON 过滤也无法完成用户要的统计。合法的 list 唯一用途:
--modality 做全量穷举/审计(不是为了"找最强")total_count 这种统计--name foo 精确匹配(agent 几乎用不到,因为 search 也能命中)models list 替代references/arkcli-models-list.mdreferences/arkcli-models-get.mdreferences/arkcli-models-search.mdreferences/arkcli-models-scenario-table.md -- 场景化推荐表(意图排序最高权重)+ JOIN 验证协议references/arkcli-models-activate.md© volcengine, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 7 other files (references) in skills/arkcli-models of volcengine/ark-cli.
Open the folder on GitHubat commit fb5b7be
Arkcli Models next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Arkcli Models this skillvolcengine/ark-cli | 140 | — | ~4.5k | Automated safety check: Pass | Apache-2.0 | |
| Agentselevenlabs/skills | 479 | — | ~6.5k | Automated safety check: Pass | MIT | |
| Int Evolution Goevolution-foundation/evo-nexus | 544 | — | ~1.5k | Automated safety check: Pass | Custom licence | |
| Telnyx Voice Media Curlteam-telnyx/ai | 220 | — | ~4.6k | Automated safety check: Pass | MIT | |
| API DesignerJeffallan/claude-skills | 12k | 2 repos | ~2k | Automated safety check: Pass | MIT | |
| Paperclippaperclipai/paperclip | 98k | — | ~9.6k | Automated safety check: Pass | MIT |
elevenlabs/skills
Build voice AI agents with ElevenLabs. An agent skill from elevenlabs/skills.
evolution-foundation/evo-nexus
Manage Evolution Go instances and send WhatsApp messages via the Evolution Go REST API.
team-telnyx/ai
Play audio files, use text-to-speech, and record calls. An agent skill from team-telnyx/ai.
Jeffallan/claude-skills
Designs REST and GraphQL APIs from resource modeling to an OpenAPI 3.1 contract, with versioning, pagination and RFC 7807 error handling.
paperclipai/paperclip
Interact with the Paperclip control plane API for task coordination and governance.
ever-works/ever-works
Build production-ready Node.js backend services with Express/Fastify, implementing middleware patterns, error handling, authentication, database integration, and API design best practices.
volcengine/ark-cli
Inspect or invoke locally registered ArkCLI actions when product commands cannot cover a task.
volcengine/ark-cli
arkcli +code-example:为指定基础模型生成多语言(Python / Go / Java / Node / curl)调用示例代码并写入本地文件。数据源是火山方舟 OpenTOP OpenGetSampleCode。当用户需要拿某个基础模型的 SDK / curl 调用示例、保存为本地接入模板时使用。反触发:TTS/ASR/语音模型没有 arkcli…
volcengine/ark-cli
arkcli 本地配置管理。处理 profile 配置归因、update.mode 的 automatic/disabled 策略、config reset 与历史 yaml 排障;profile 类操作优先使用 arkcli profile <subcmd。
volcengine/ark-cli
arkcli 自定义模型仓库管理:从 TOS 导入自定义模型、查询/筛选自定义模型、查看详情、改名、删除、查询可用量化模式、量化已就绪的模型。任何提到自定义模型 ID(cm-)的管理、部署准备,或要求用 cm- 直接对话/推理/试效果的边界判断,都必须使用本 skill。注意:查询火山公共基础模型(doubao 等 foundation models)走 arkcli-models;本…
volcengine/ark-cli
arkcli +deploy:普通创建推理接入点(Endpoint)的统一首选入口。用户说『创建/新建/create 一个 endpoint/接入点』或『部署/上线/deploy 某模型』时优先走这里;但脚本化 / CI / 无护栏 / 原始 raw CRUD 创建是唯一例外,必须改走 arkcli-infer-endpoint,不能由本 skill…
volcengine/ark-cli
检索、读取与总结方舟官方文档。用户给出 ark.volcengine.com 文档 URL 或 /docs/ 路径、要求读链接、官方说明、API 契约或必填字段,询问 CC Switch 等第三方客户端的方舟图形配置流程,以及官方网页读取失败时使用。不用于业务调用、资源操作、CLI 帮助或通用知识。
Categories
arkcli 模型查询与基础模型服务激活能力:列出、搜索、获取火山公共基础模型详情,以及用户明确要求的开通/激活模型服务(arkcli models activate);Volc 还支持 TTFT/TPOT 性能排名、延迟趋势和输入长度对比。激活已有基础模型服务不等于部署/创建 Endpoint,不得转成 +deploy。优先使用产品命令 arkcli models ...,而不是直接调用…. Arkcli Models is an agent skill from volcengine/ark-cli.,而不是直接调用 Raw API。反触发:用户的最终目标是创建 / 部署 Endpoint 时,本 skill 承担有界的只读候选查询,owning skill 由创建路径确定:普通产品创建走 arkcli-deploy;用户显式要求 raw CRUD / CI / 无守卫的 infer endpoint create 走 arkcli-infer-endpoint。候选首先执行 models search ...
Arkcli Models fits situations like: tasks that involve REST APIs.
Run `npx skills add volcengine/ark-cli --skill arkcli-models -a claude-code`. Or copy the skill folder (skills/arkcli-models in volcengine/ark-cli) into .claude/skills/arkcli-models in your project. Claude Code loads it when a task matches its description.
Run `npx skills add volcengine/ark-cli --skill arkcli-models -a codex`. Or copy the skill folder (skills/arkcli-models in volcengine/ark-cli) into .agents/skills/arkcli-models in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add volcengine/ark-cli --skill arkcli-models -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/arkcli-models, .gemini/skills/arkcli-models, .github/skills/arkcli-models and .opencode/skills/arkcli-models in your project.
SKILL.md names no scripts, command-line tools or credentials: Arkcli Models is instructions for the agent only.
SKILL.md names 1 domain. As links in the text: ark.volcengine.com. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
Arkcli Models is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
About 4.5k tokens (SKILL.md is roughly 18k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 13k tokens, read only when the agent opens those files.
Skills that share tags, products or a category with Arkcli Models: Agents (elevenlabs/skills, 479 stars), Int Evolution Go (evolution-foundation/evo-nexus, 544 stars), Telnyx Voice Media Curl (team-telnyx/ai, 220 stars) and API Designer (Jeffallan/claude-skills, 12k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
volcengine (a GitHub organization) maintains it in volcengine/ark-cli, which has 140 GitHub stars. The repository holds 25 skills in this directory. The repository was last updated on September 29, 2026.
Source: volcengine/ark-cli on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.