SearchCLI CLI
SearchCLI is a Volcengine AI Search tool that evaluates results against test queries and retains an experiment plan and report.
- 1,191 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The CLI uploads data and changes AI Search resources with supplied credentials; optional LLM tasks send material to the chosen model.
Install
Manual install
git clone git@github.com:volcengine/SearchCLI.git vs
cd vs
bash ./scripts/install.shFirst full README installation block for users and agents.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The CLI uploads data and changes AI Search resources with supplied credentials; optional LLM tasks send material to the chosen model.
This is third-party code. Review the repository files before installing.
What it does
SearchCLI is a command-line tool that manages data, apps and retrieval workflows in Volcengine AI Search. Users upload data, inspect inferred schemas and attach datasets to applications before testing search, recommendations or conversational retrieval. Tuning includes query generation, experiment plans and reports. Installable Viking skills guide external agents through previews, confirmations and read-after-write verification.
Who it is for. Volcengine search and recommendation builders needing reviewable data and retrieval-quality workflows.
Good fit when
- A product catalog is uploaded to a new dataset for its owner to review the schema.
- After data is connected, the team inspects actual search responses.
- Retrieval quality is evaluated against prepared queries.
- An agent configures an application and presents a plan before writing changes.
Not a fit when
- You need a standalone engine independent of Volcengine.
- Data must not reach the selected cloud service.
- You need public web search rather than application-data retrieval.
Example request
Prepare a test JSONL import into a new AI Search dataset. Show the schema and write plan, then wait for approval.Limitations
The setup needs Node.js 20 or later, Git and Volcengine AK/SK credentials with AI Search access. The first installation method clones over SSH, so GitHub SSH access must be configured. Model-based query generation and relevance evaluation need a separate OpenAI-compatible API and key. Data and configuration are written to a cloud application. Review the dataset, schema and target resource identifiers before writing.
How to disable. Stop workflows, remove agent skills and revoke unused keys. Cloud apps and datasets remain in Volcengine and need separate management.
FAQ
How do I check authorization?
Run `vs auth status --json` and `vs doctor --json` before onboarding.
Can I import environment credentials?
Yes. Use `vs auth import-env` for `VIKING_AK` and `VIKING_SK`. Keep their values out of terminal output and reports.
Why is LLM login separate?
Separate model access supports query generation and relevance evaluation. `vs llm status --json` shows that connection’s status before an experiment.
Must onboarding create an app?
No. The workflow can stop after the dataset is written. Create and connect an application later when moving on to retrieval scenarios.
How does an agent get guidance?
Install Viking skills after the CLI. `vs skill list` shows the available guidance skills for the agent.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI agent for quantitative investment research: it reads broker reports, ideates factors and computes IC on offline data.
The CLI imports CSV and JSON into SQLite, runs SQL and reshapes tables for agent-assisted local data preparation.
The CLI selects and changes JSON, YAML and other structured values so agents can transform data with one query language.
An MCP server and skill for Sciverse paper search: metadata, semantic retrieval, citations and full text for agents.
For a repeatable experiment, an agent downloads a Kaggle dataset to a working directory and retrieves notebook outputs.
A skill retrieves Carrel evidence, reads source passages and images and helps agents cite original files in answers.
README in short
Public command groups separate authentication, datasets, apps and search. Additional commands handle recommendations and conversational retrieval.