# OpenRouter, Vercel AI Gateway,# OpenCode Go, DeepSeek, Cloudflare,# Google AI Studio, Kilo Gateway,# Amazon Bedrock,# OAuth: Codex, SuperGrok, GitHub Copilot, Cursor# Local: Ollama, LM Studio, llama.cpp# more...aivo keys add
导出与导入
通过一个密码加密文件,在不同设备之间迁移 API Key。
# prompts for a passwordaivo keys export ~/keys.bak
# same password on the other machine, or pass a URLaivo keys import ~/keys.bak
aivo keys import https://example.com/keys.bak
# non-interactive (CI / scripts)aivo keys export ~/keys.bak --password-stdin<<<"my secret"
其他命令
列出、启用、查看、编辑、连通性探测已保存的 API Key,或重新认证 OAuth。
# list all saved keysaivo keys
# quickly add a key with optionsaivo keys add --base-url https://openrouter.ai/api --key sk-xxx
# activate a key to useaivo keys use
aivo keys use mykey
# print saved data of a keyaivo keys cat
# edit a saved keyaivo keys edit
# health-checkaivo keys ping
# OAuth re-login / rotate a keyaivo keys reauth
# for more optionsaivo keys --help
使用 -k 指定已保存的 API Key;省略值可打开 API Key 选择器。同时使用不带值的 -k 和 -m,可以一次选择两者。
aivo claude -k openrouter
aivo claude -k copilot
aivo claude -k
使用 -m 指定模型;省略值可打开模型选择器。应用模型后,aivo claude 会在下次运行时沿用该模型。
aivo claude -m moonshotai/kimi-k2.5
aivo claude -m
各槽位独立模型 Claude Code 支持同会话多模型编排,涵盖 haiku/sonnet/opus/fable 槽位及 Subagent 覆盖项。每个槽位均可独立指定模型;参数缺省时自动弹出选择器。
aivo claude --sonnet-model deepseek-v4-pro --haiku-model deepseek-v4-flash
aivo claude --fable-model gpt-5.4 --subagent-model claude-haiku-4-5
# open the picker for one slotaivo claude --opus-model
各模式独立模型Amp 可在 rush、smart、deep 和 large 模式之间切换。每种模式都能指定不同的模型;模式参数不带值时,会打开对应的模型选择器。
aivo amp --smart-model claude-sonnet-4.6 --rush-model claude-haiku-4-5
aivo amp --deep-model gpt-5.4-thinking --large-model deepseek-v4-pro
# pin the starting mode, or omit the value to open the mode pickeraivo amp --mode deep
aivo amp --mode
不指定工具名时,aivo run 会打开 Coding Agent 和插件的交互式选择器,默认选中上次使用的工具。
aivo run
启动别名
为 Coding Agent 及其常用参数创建启动别名,之后可通过 aivo <name>(或完整形式 aivo run <name>)一键启动。运行时传入的参数会覆盖已保存的参数,轻松复用槽位模型配置。
# Claude Code with separate models per slotaivo alias work claude -k openrouter \--opus-model deepseek/deepseek-v4-pro \--sonnet-model deepseek/deepseek-v4-flash \--haiku-model deepseek/deepseek-v4-flash \# Amp with separate models per mode, pinned to deep modeaivo alias task amp -k openrouter \--smart-model openai/gpt-5.5 \--deep-model anthropic/claude-opus-4.7 \--rush-model openai/gpt-5.4 \--large-model deepseek/deepseek-v4-pro
# run themaivo work
aivo task
# start the interactive coding agentaivo code
# start with a task, key, or modelaivo code "fix the failing login test"aivo code -k openrouter -m anthropic/claude-sonnet-4.6
# resume your latest sessionaivo code --resume last
规划、执行与恢复
Plan 模式会先以只读方式调查,待你批准方案后再实现。Goal 模式会围绕目标自主持续工作;Rewind 可将对话和文件修改一起恢复到更早的轮次。
# investigate read-only, then approve the plan to build/plan redesign the authentication flow
# keep working autonomously until the objective is done/goal make the test suite pass
# restore the project to an earlier turn/rewind
# full agent loop: inspect, edit, run, and verifyaivo code -e"fix the failing tests"# bound unattended work and return a scriptable resultaivo code -e"audit dependencies"\--max-steps 40 --max-cost 2 \--output-format json
# continue a prior headless runaivo code -e"now fix what you found"--resume last
恢复、附加与分享
继续已保存的会话、附加文本或图片供 Agent 读取,或创建实时脱敏的会话查看 URL。
# resume a past sessionaivo code --resume# attach files or images for the agent to readaivo code --attach README.md --attach screenshot.png
# share a live, redacted view of the sessionaivo code --share
aivo alias fast=claude-haiku-4-5
aivo alias mimo xiaomi/mimo-v2-pro
# use itaivo claude -m fast
aivo code -k vercel -m mimo
# list and removeaivo alias
aivo alias rm fast
aivo hf:Qwen/Qwen2.5-0.5B-Instruct-GGUF
aivo https://huggingface.co/allenai/Olmo-3-1025-7B
# a local .gguf works tooaivo code -m ./Qwen2.5-0.5B-Instruct-Q4_K_M.gguf
AIVO_LLAMA_CTX=16384# smaller context on a low-RAM machineAIVO_LLAMA_NGL=20# GPU layers (AIVO_GPU=cpu to stay on CPU)AIVO_LLAMA_ARGS='--temp 0.1'# extra llama-server flagsAIVO_LLAMA_MMPROJ=off# skip the vision projectorAIVO_LLAMA_DRAFT=off# skip the draft model
混用本地与远程
凡接受 -m/--model 处均可传入 hf:。配合 Claude Code 与 Amp 的槽位参数,可在同一次会话中混用本地与远程模型。
aivo claude -m hf:Qwen/Qwen2.5-0.5B-Instruct-GGUF
aivo pi -m hf:Qwen/Qwen2.5-0.5B-Instruct-GGUF
# mix: remote opus, local sonnetaivo claude \--opus-model claude-opus-4.7 \--sonnet-model hf:Qwen/Qwen2.5-0.5B-Instruct-GGUF
重新打开已缓存的模型
单独传入 hf: 即可唤起本地缓存交互选择器,无需重复输入完整的仓库路径。
aivo code hf:
aivo claude hf:
管理本地缓存
aivo hf 列出所有已缓存仓库;--verbose 可展开查看各仓库下下载的具体量化版本。
# list cached reposaivo hf
# pre-pull a model (handy before going offline)aivo hf pull hf:Qwen/Qwen2.5-0.5B-Instruct-GGUF
# delete one quant or a whole repoaivo hf rm <repo>--quantQ5_K_Maivo hf rm <repo>--all-y# wipe the whole cacheaivo hf clean -y
# pick a session in the current projectaivo logs share
# share a specific row by id prefixaivo logs share 1335c631
# pick from every project on this machineaivo logs share --all# follow updates and open in the browseraivo logs share --open# skip redaction (be sure)aivo logs share --no-redact
用量统计
直接解析各工具的原生会话数据,统一汇总并呈现跨 Agent、跨模型的 Token 消耗看板。
aivo stats
aivo stats claude
aivo stats code
筛选与输出格式
支持输出原始数值、按服务商筛选、设置时间范围,以及绕过缓存,便于脚本处理。
# raw numbers for scriptsaivo stats -n# filter by provideraivo stats -s openrouter
# last N units (m, h, d, w)aivo stats --since 7d
aivo stats claude --since 24h
# show all modelsaivo stats -a# bypass cacheaivo stats -r
aivo update
# force update even if installed via a package manageraivo update --force# restore the previous version from the last backupaivo update --rollback