SearchCLI is a command-line interface for connecting agent or business systems to Volcengine’s search, recommendation, and conversational-retrieval services. It is for developers who need to onboard data, build and test retrieval workflows, adjust search strategies, inspect poor results, and verify changes. Its catalogue entries provide installable skills for operating these workflows.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add volcengine/SearchCLI --skill vs-search-tuninggit clone --depth 1 https://github.com/volcengine/SearchCLIWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/volcengine/searchcli/vs-search-tuning)<a href="https://agentmods.dev/skills/volcengine/searchcli/vs-search-tuning"><img src="https://agentmods.dev/badge/skills/volcengine/searchcli/vs-search-tuning/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/volcengine/searchcli/vs-search-tuning"><img src="https://agentmods.dev/badge/skills/volcengine/searchcli/vs-search-tuning.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 115 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00029 | $0.03268 |
| Opus 5.5 | $0.00012 | $0.01307 |
| Sonnet 5.5 | $0.00006 | $0.00654 |
| Haiku 4.5 | $0.00003 | $0.00327 |
Grade A, and why
vs-search-tuning scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Viking Search Tuning
When to Use
Use this skill when the user wants an external agent to evaluate and tune text search similarity for an existing AI Search application and dataset.
This first version is for similarity tuning. It defaults to text-query/text-item judging, fixes mode=UserDefined, and tunes the user-defined recall strategy, recall weights, keyword match ratio, and max retrieved count. It can optionally use text+image LLM judging when visual relevance matters; image fields are taken only from GetAppDataConfig.ImageIndexFields. It does not tune rerank, personalization, hotness, boost/bury, sort rules, serving controls, or business operating rules.
Version Check
Before starting this skill workflow, run vs version check --json. Continue only when status is up-to-date. If status is update-available, stop and tell the user to update the cloned vs repository, then run git pull --ff-only, bash ./scripts/install.sh, and bash ./scripts/install-skills.sh all --target auto --force (PowerShell: scripts/install.ps1 and scripts/install-skills.ps1). If the status is unknown, stop and report that the CLI version could not be verified.
Preconditions
- an
application-idis available - a
dataset-idis preferred; if omitted, the CLI can try to infer a unique search dataset from the application - Viking auth is configured with
vs auth status - LLM config is available through
vs llm login,vs llm import-env, orVIKING_LLM_BASE_URL/VIKING_LLM_API_KEY/VIKING_LLM_MODELwhen generating queries or using LLM relevance labels - a query file with
sourceItemIdscan be evaluated with--label-source source-itemfor a fast first-pass silver-label run without LLM relevance judging - the user understands that LLM relevance labels are silver labels and should be reviewed before high-risk production changes
Commands
llm login/llm import-env/llm status: configure and verify OpenAI-compatible LLM credentials without exposing API keys in chat or plain configsearch tune llm-check: verify CLI-managed LLM configurationsearch tune validate: validate a query set locally before planning or running; reports schema issues, duplicate ids/text, sourceItemIds coverage, query type skew, and a label-source recommendationsearch tune query-generate: generate a reusable synthetic query set from paged dataset samples with batched concurrent LLM calls when the user has no query set; add--retrievable-field-onlywhen the user wants generation constrained to textIndexFieldsfrom app dataset config, excludingImageIndexFieldssearch tune plan: show query source, candidate strategies, estimated requests/labels, parameter coverage, source-item coverage, warnings, and suggested first-pass size before runningsearch tune run: generate or load queries, run candidate search strategies, label top results, compute metrics, and write artifacts; supports--label-source llm|source-item|auto,--judge-input text|text-image,--max-judge-images,--llm-retries,--max-label-failure-rate, and--verbose; use--resume-run-id <run-id>to continue an interrupted runsearch tune report: read a previous tuning reportsearch tune compare: compare completed tuning runs with--run-ids, or compare existing scenes online with--scene-ids --queriesusing source-item silver labelssearch tune apply: create a new candidate search scene from a completed tuning report recommendationapp status/doctor: verify app and local environment readiness
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed · +4 lines 32532352805b
- 1mo ago First seen · 125 lines · 29 tokens per session scan A a9955c045fec
vs-search-tuning is a skill published in the GitHub repository volcengine/SearchCLI (1,189 stars, last pushed 5d ago), licensed Apache-2.0. It adds 29 tokens to every session and 3,268 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
geo-readiness
Make a site citable by AI engines (ChatGPT, Perplexity, Google AI Overviews) — Generative Engine Optimization. Audits robots.txt AI-crawler policy and the llms.txt layer, generates llms.txt / llms-full.txt from a static build or sitemap, and plans Query Fan-out content (the sub-queries AI engines spawn from a seed…
thegraph-mcp-skill
Use The Graph Subgraph MCP through UXC via native SSE with a fixed linked command for subgraph discovery, schema retrieval, deployment selection, and GraphQL query execution with help-first inspection and explicit auth handling.
qmd-mcp-skill
Use a local QMD knowledge base through UXC over MCP stdio, with daemon-backed session reuse and typed retrieval flows that avoid repeated model warmup and unnecessary query-expansion latency.
youtube-content
YouTube Data API — search, video details, channel info, transcripts.
prompt-engineer
Turn vague instructions into precision-crafted prompts and system messages that get the best out of any LLM.
shap
Model interpretability and explainability using SHAP (SHapley Additive exPlanations). Use this skill when explaining machine learning model predictions, computing feature importance, generating SHAP plots (waterfall, beeswarm, bar, scatter, force, heatmap), debugging models, analyzing model bias or fairness, comparing…