メニュー
Grounded semantic search and asynchronous research through Centra AI's router-owned Exa MCP. Use for controlled web evidence, company research, known-URL extraction, outbound briefs, or source-grounded drafts.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/Sidiora-Labs/centra-llm-agents --skill exa-searchコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Design, implement, and audit inclusive digital products using WCAG 2.2 Level AA standards. Use this skill to generate semantic ARIA for Web and accessibility traits for Web and Native platforms (iOS/Android).
Head-to-head comparison of coding agents (Claude Code, Aider, Codex, etc.) on custom tasks with pass rate, cost, time, and consistency metrics
Design and optimize AI agent action spaces, tool definitions, and observation formatting for higher completion rates.
SKILL.md を表示中
| name | exa-search |
| description | Grounded semantic search and asynchronous research through Centra AI's router-owned Exa MCP. Use for controlled web evidence, company research, known-URL extraction, outbound briefs, or source-grounded drafts. |
| origin | Centra AI/Paxlabs |
Use Centra AI's router-owned Exa lane for semantic retrieval, extractive contents, and bounded multi-source research. Neo never receives the vendor API key.
Use only the shipped exa__* tools. The Centra router owns vendor credentials, user scoping, rate and spend limits, caching, run ownership, metering, and cancellation. Never ask the user for an Exa key, install a vendor MCP, or call Exa directly.
web_search / web_news plus fetch as the cheap default for ordinary discovery.exa_search when semantic matching, source/date controls, financial reports, or deeper synthesis materially improves retrieval.exa_contents for extractive evidence or bounded full text from 1-10 known URLs.exa_searchSupports query, type, num_results, category, domain include/exclude lists, publication-date bounds, and an extractive highlight_query.
exa_search(query: "current WebAssembly component model adoption", num_results: 8, start_published_date: "2026-01-01")
exa_contentsReturns highlights or bounded full text and a status for every requested URL. An HTTP-successful batch can still be partial; preserve and report every failed URL status.
exa_contents(urls: ["https://example.com/report"], highlight_query: "revenue guidance")
Call exa_research_start, then poll the owned run with exa_research_get. Only completed terminal output and its grounding are authoritative. Use exa_research_continue for a bounded follow-up and exa_research_cancel when queued or running work is no longer needed.
exa_outbound_brief returns sourced company/person context and talking points; it sends nothing.exa_social_draft returns a source-grounded draft; it publishes nothing.deep-research for a complete multi-source workflow.market-research for decision-oriented business analysis.