[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-76683":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":16,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":10,"pushedAt":10,"updatedAt":41,"readmeContent":42,"aiSummary":43,"trendingCount":15,"starSnapshotCount":15,"syncStatus":44,"lastSyncTime":45,"discoverSource":46},76683,"SearchCLI","volcengine\u002FSearchCLI","volcengine","Open CLI for integrating AI search, recommendation, and conversational retrieval into agent systems and business systems","",null,"TypeScript",453,22,13,0,102,406,306,94.09,"Apache License 2.0",false,"main",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,7],"agent-tools","ai-agents","ai-search","ai-search-engine","automation","cli","command-line-tool","conversational-search","information-retrieval","rag","recommendation","recommendation-system","retrieval","search-cli","search-engine","semantic-search","typescript","2026-06-12 04:01:21","\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fsearchcli-logo.svg\" alt=\"SearchCLI logo\" width=\"560\" \u002F>\n\u003C\u002Fp>\n\n# SearchCLI\n\nEnglish | [中文README](README.zh_CN.md)\n\nConnect stable, tunable search, recommendation, and conversational retrieval to your agent system or business system.\n\n[Quick Start](#quick-start-human-users) · [AI Agent Setup](#quick-start-ai-agents) · [Full Agent Guide](docs\u002Fagent-quick-start.md) · [Contributing](CONTRIBUTING.md) · [Security](SECURITY.md)\n\nSearchCLI is the open CLI for AI Search on Volcengine.\n\nIf your agent system or business system needs stable, tunable information distribution services, SearchCLI gives you a practical path to integrate production-grade search, recommendation, and conversational retrieval into real workflows.\n\nWith SearchCLI and its installable `Viking skills`, external agents can onboard data, build and validate search and recommendation flows, run conversational retrieval, tune strategy configuration, inspect bad cases, and iterate on retrieval quality in a stable, reviewable way.\n\n## What SearchCLI Is\n\n- The command-line integration surface for AI Search on Volcengine.\n- A stable path for external systems to access search, recommendation, and conversational retrieval capabilities.\n- An agent-friendly workflow layer built around installable skills and automation-safe command output.\n- A reviewable execution model with dry-runs, confirmation gates, and read-after-write verification.\n\n## Who It Is For\n\n- Developers integrating AI-powered information distribution into business systems.\n- Teams building agent systems that need stable, configurable search, recommendation, and retrieval workflows.\n- Operators and solution teams who need a reviewable way to onboard data, configure applications, and verify runtime behavior before production use.\n\n## What It Enables\n\n- Item and catalog search on top of structured business data.\n- Recommendation flows connected to application scenes and user behavior.\n- Conversational retrieval experiences grounded in application search.\n- Agent workflows that can onboard data, configure applications, and validate runtime behavior with explicit review steps.\n\n## Core Capabilities\n\n- `vs item profile | plan | apply` for structured item onboarding.\n- `vs app`, `vs dataset`, and `vs data` for application and dataset management.\n- `vs search run`, `vs recommend run`, and `vs chat run` for runtime verification.\n- `vs search tune query-generate | plan | run | report` for first-version automated text-similarity evaluation and tuning.\n- Installable `Viking skills` so external agents can use the same workflows.\n\n## Requirements\n\n- Node.js 20 or newer\n- `git`\n- Volcengine AK\u002FSK with access to AI Search\n\n## Quick Start (Human Users)\n\n### 1. Install\n\n```bash\ngit clone git@github.com:volcengine\u002FSearchCLI.git vs\ncd vs\nbash .\u002Fscripts\u002Finstall.sh\n```\n\n### 2. Authenticate\n\nIf the current shell already has `VIKING_AK` and `VIKING_SK`:\n\n```bash\nvs auth import-env\nvs auth status --json\nvs doctor --json\n```\n\nOtherwise, run interactive login in a real terminal:\n\n```bash\nvs auth login\n```\n\nIf you will use search tuning query generation or LLM relevance judging, configure an OpenAI-compatible LLM API without placing the API key in plain config:\n\n```bash\nvs llm login\nvs llm status --json\nvs search tune llm-check --live --json\n```\n\nIf the current shell already has `VIKING_LLM_BASE_URL`, `VIKING_LLM_API_KEY`, and `VIKING_LLM_MODEL`, use `vs llm import-env` instead. The API key is stored in the local secure credential store; base URL and model are stored as non-secret config.\n\n### 3. Run the First Onboarding Flow\n\nIf the user wants a new app plus bind-time config review and runtime verification, use the `dataset+app` path:\n\n```bash\nvs item profile --file .\u002Fitems.json --pretty\nvs item plan --file .\u002Fitems.json --goal \"Build item search\"\nvs item apply --plan-dir .\u002F.viking\u002Fitem-plans\u002F\u003Cplan> --dry-run\nvs item apply --plan-dir .\u002F.viking\u002Fitem-plans\u002F\u003Cplan> --confirm-review --wait-ready --run-trials\n```\n\nIf you only need dataset provisioning, use the `dataset-only` path, generate a dataset-only plan with `--skip-app`, and stop after dataset create + ingest:\n\n```bash\nvs item profile --file .\u002Fitems.json --pretty\nvs item plan --file .\u002Fitems.json --goal \"Build item search\" --skip-app\nvs dataset create --data @dataset-create.json\nvs dataset ingest --dataset-id \u003Cdataset-id> --fields @\u003Cnormalized-items-artifact>\n```\n\nPrefer `dataset-create.json` when the plan emitted it so dataset creation keeps `Schema` and `DataFieldConfig` together. The `--name \u003Cdataset-name> --type item --schema @schema.json` form remains the manual schema-only fallback when a full create payload is unavailable or unsuitable.\n\n`--skip-app` is also accepted by `vs item provision` and `vs item apply` as an execution-time guard rail when you need to enforce the dataset-only boundary from an existing plan.\n\nIf you need a video dataset, do not rely on the default type. Always pass `--type video` explicitly:\n\nFor `dataset+app`:\n\n```bash\nvs item profile --file .\u002Fvideos.jsonl --type video --pretty\nvs item plan --file .\u002Fvideos.jsonl --type video --goal \"Build video search\"\nvs item apply --plan-dir .\u002F.viking\u002Fitem-plans\u002F\u003Cplan> --dry-run\nvs item apply --plan-dir .\u002F.viking\u002Fitem-plans\u002F\u003Cplan> --confirm-review --wait-ready --run-trials\n```\n\nFor `dataset-only`:\n\n```bash\nvs item profile --file .\u002Fvideos.jsonl --type video --pretty\nvs item plan --file .\u002Fvideos.jsonl --type video --goal \"Build video search\" --skip-app\nvs dataset create --data @dataset-create.json\nvs dataset ingest --dataset-id \u003Cdataset-id> --fields @\u003Cnormalized-items-artifact>\n```\n\nFor video dataset-only provisioning, prefer `dataset-create.json` so the create request includes `DataFieldConfig`; `--schema @schema.json` alone can fail with `MissingParameter.DefaultFieldStrategy`.\n\n## Quick Start (AI Agents)\n\nIf an external agent needs to operate AI Search through this repository:\n\n### 1. Install SearchCLI\n\n```bash\ngit clone git@github.com:volcengine\u002FSearchCLI.git vs\ncd vs\nbash .\u002Fscripts\u002Finstall.sh\n```\n\n### 2. Install Viking skills\n\n```bash\nnpx skills add \"git@github.com:volcengine\u002FSearchCLI.git\" -y -g\n```\n\nThe default public skill bundle is:\n\n- `vs-shared`\n- `vs-item-onboarding`\n- `vs-search`\n- `vs-search-tuning`\n- `vs-chat`\n- `vs-recommend`\n\n### 3. Authenticate\n\nIf the current shell already has `VIKING_AK` and `VIKING_SK`, prefer:\n\n```bash\nvs auth import-env\n```\n\nOtherwise:\n\n```bash\nvs auth login\n```\n\n### 4. Verify\n\n```bash\nvs --help\nvs auth status --json\nvs llm status --json\nvs doctor --json\nvs skill list\n```\n\n## Public Command Groups\n\n- `vs auth`\n- `vs llm`\n- `vs doctor`\n- `vs skill`\n- `vs item`\n- `vs app`\n- `vs dataset`\n- `vs data`\n- `vs search`\n- `vs chat`\n- `vs recommend`\n\n## Documentation\n\n- [Agent Quick Start](docs\u002Fagent-quick-start.md)\n- [Contributing](CONTRIBUTING.md)\n- [Security](SECURITY.md)\n\n## Maintainer Workflow\n\nIf you are maintaining the open-source repository itself, the local skill tooling is:\n\n```bash\nvs skill list\nvs skill init viking-demo-skill\nvs skill validate\nvs skill install all\n```\n\nBuild and run repository checks:\n\n```bash\nnpm install\nnpm run validate:skills\nnpm run build\nnpm run test:acceptance:dist\n```\n\n## Contribution\n\nPlease check [Contributing](CONTRIBUTING.md) for more details.\n\nExternal contributors must complete the Contributor License Agreement (CLA) before a pull request can be accepted.\n\n## Code of Conduct\n\nPlease check [Code of Conduct](CODE_OF_CONDUCT.md) for more details.\n\n## Security\n\nIf you discover a potential security issue in this project, or think you may have discovered a security issue, we ask that you notify Bytedance Security via our [security center](https:\u002F\u002Fsecurity.bytedance.com\u002Fsrc) or [vulnerability reporting email](mailto:sec@bytedance.com).\n\nPlease do **not** create a public GitHub issue.\n\n## License\n\nThis project is licensed under the [Apache-2.0 License](LICENSE).\n","SearchCLI 是一个用于将AI搜索、推荐和对话检索功能集成到代理系统和业务系统的开源命令行工具。它采用TypeScript编写，提供了一系列稳定且可调的信息分发服务，支持结构化数据导入、应用及数据集管理、运行时验证等功能，并通过可安装的“Viking技能”来增强外部代理系统的集成能力。适用于需要将智能信息处理能力融入实际工作流中的开发者、构建稳定配置搜索与推荐流程的团队以及希望以可审查方式上线数据、配置应用并验证运行行为的操作员或解决方案团队。该项目遵循Apache License 2.0许可协议，当前在GitHub上获得了99个星标。",2,"2026-06-11 03:55:05","CREATED_QUERY"]