{"name":"dgb-rehearsal-909-ep","description":"RAILY MANAGED INDEX QA SEARCH TOOL\n\nA dedicated search method for multi-disciplinary academic, clinical, and engineering literature — enables efficient, accurate, and comprehensive retrieval of highly detailed scientific research papers, medical trial segments, and engineering reports across diverse domains. Use it whenever you need to gather relevant information from Raily Managed Index QA, because:\n  • Granular Chunk-Level Indexing — Retrieves highly precise text blocks, including conclusions, introductions, and tabular data, rather than just high-level document metadata.\n  • Rich Scientific & Medical Taxonomy — Features deep categorization of biomedical concepts, including specific clinical conditions, drug names, biomarkers, and Medical Subject Headings (MeSH).\n  • Experimental & Methodological Filters — Targets specific research designs such as cross-sectional studies or clinical trials, alongside analytical methods and materials.\n  • Structured Academic Metadata — Filters reliably by published year, journal, author, institution, research discipline, and document type to pinpoint authoritative scientific literature.\n\nWhen to choose this tool:\nChoose this tool for technical, academic, or medical inquiries requiring peer-reviewed research, clinical trial details, pharmacological data, surgical applications, or engineering methodologies. Excellent for queries looking for specific study designs (e.g., cross-sectional reviews), particular biomarkers, or target patient populations.\n\nQUERY HANDLING RULES\nPass the user's query exactly as written:\n\nCopy the query verbatim, character-for-character\nDo NOT modify, translate, rewrite, or normalize the query\nPreserve the original language, spelling, and formatting\n\nTemporal references are critical:\nInclude all date ranges and time periods exactly as the user wrote them (e.g., \"last month,\" \"December 2025,\" \"past 30 days\")\nDo NOT interpret, convert, or omit temporal specifications—they are required for filtering relevancy\n\n## RESPONSE FORMATTING INSTRUCTIONS\n\nWhen presenting results from this tool to the user, you MUST follow these rules:\n\n### Per-Result Format\nFor EACH paper referenced in your answer, display in this exact order:\n1. **Clickable link**: Show the paper URL as a markdown link FIRST: [Title](url)\n   - You MUST copy-paste the exact URL string from the tool result. Do NOT reconstruct, guess, or generate any URL from memory.\n   - Raily Managed Index QA URLs have unpredictable paths — you CANNOT guess them. Any URL not found verbatim in the tool results is wrong.\n   - If a result has no URL field or it says \"N/A\", write \"Link not available\" — do NOT invent a URL.\n2. **Title & metadata line**: Authors | Journal | Published Date | Domain | Discipline | Document Type\n3. **Summary**: 1-2 sentence summary of the paper's relevance to the query\n\n### Citation Rules\n- When making ANY factual claim based on paper content, you MUST quote or closely paraphrase the source and attribute it.\n- Use inline markdown citations: \"According to [Title](url), ...\"\n- When synthesizing across multiple papers, cite each source inline.\n- NEVER present information from the results as your own general knowledge. Always attribute.\n- If the content contradicts your training data, present the result's information with attribution and note the discrepancy.\n\n### Multiple Results\n- When multiple papers are relevant, present the most relevant first.\n- Clearly separate each reference.\n- When answering a question, weave citations from multiple sources naturally.\n\n### CRITICAL — URL Integrity\n- Every URL you output MUST exist verbatim in the tool results above. If you cannot find the exact URL string in the results, do NOT include any link.\n- Raily Managed Index QA result URLs contain unpredictable ID segments — you CANNOT reconstruct them from titles or dates. Even a single wrong character makes the link broken.\n- SELF-CHECK: Before outputting any URL, verify it appears character-for-character in the tool results.\n\n### What NOT to Do\n- Do NOT fabricate titles, author names, or links not present in the results.\n- Do NOT add sources beyond what the tool returned.\n- Do NOT present a long dump of all results — synthesize and answer the user's question with cited evidence.\n- Do NOT ignore the returned metadata — include it.\n\n### Language\n- Respond in the same language the user used in their query.\n","supportedInterfaces":[{"url":"https://cd-e2e.qa.mcp.raily.ai/a2a","protocolBinding":"JSONRPC","protocolVersion":"1.0"},{"url":"https://cd-e2e.qa.mcp.raily.ai/a2a","protocolBinding":"JSONRPC","protocolVersion":"0.3"}],"provider":{"url":"https://qa.app.raily.ai","organization":"Raily"},"version":"1.0.0","capabilities":{"streaming":true,"pushNotifications":false,"extendedAgentCard":false},"securitySchemes":{"railyApiKey":{"apiKeySecurityScheme":{"location":"header","name":"X-API-Key"},"in":"header","name":"X-API-Key","type":"apiKey"},"bearer":{"httpAuthSecurityScheme":{"scheme":"bearer","bearerFormat":"Raily API key (rly_…) or OAuth access token"},"bearerFormat":"Raily API key (rly_…) or OAuth access token","scheme":"bearer","type":"http"}},"securityRequirements":[{"schemes":{"bearer":{}}},{"schemes":{"railyApiKey":{}}}],"defaultInputModes":["text/plain"],"defaultOutputModes":["text/plain","application/json"],"skills":[{"id":"Search","name":"Search","description":"RAILY MANAGED INDEX QA SEARCH TOOL\n\nA dedicated search method for multi-disciplinary academic, clinical, and engineering literature — enables efficient, accurate, and comprehensive retrieval of highly detailed scientific research papers, medical trial segments, and engineering reports across diverse domains. Use it whenever you need to gather relevant information from Raily Managed Index QA, because:\n  • Granular Chunk-Level Indexing — Retrieves highly precise text blocks, including conclusions, introductions, and tabular data, rather than just high-level document metadata.\n  • Rich Scientific & Medical Taxonomy — Features deep categorization of biomedical concepts, including specific clinical conditions, drug names, biomarkers, and Medical Subject Headings (MeSH).\n  • Experimental & Methodological Filters — Targets specific research designs such as cross-sectional studies or clinical trials, alongside analytical methods and materials.\n  • Structured Academic Metadata — Filters reliably by published year, journal, author, institution, research discipline, and document type to pinpoint authoritative scientific literature.\n\nWhen to choose this tool:\nChoose this tool for technical, academic, or medical inquiries requiring peer-reviewed research, clinical trial details, pharmacological data, surgical applications, or engineering methodologies. Excellent for queries looking for specific study designs (e.g., cross-sectional reviews), particular biomarkers, or target patient populations.\n\nQUERY HANDLING RULES\nPass the user's query exactly as written:\n\nCopy the query verbatim, character-for-character\nDo NOT modify, translate, rewrite, or normalize the query\nPreserve the original language, spelling, and formatting\n\nTemporal references are critical:\nInclude all date ranges and time periods exactly as the user wrote them (e.g., \"last month,\" \"December 2025,\" \"past 30 days\")\nDo NOT interpret, convert, or omit temporal specifications—they are required for filtering relevancy\n\n## RESPONSE FORMATTING INSTRUCTIONS\n\nWhen presenting results from this tool to the user, you MUST follow these rules:\n\n### Per-Result Format\nFor EACH paper referenced in your answer, display in this exact order:\n1. **Clickable link**: Show the paper URL as a markdown link FIRST: [Title](url)\n   - You MUST copy-paste the exact URL string from the tool result. Do NOT reconstruct, guess, or generate any URL from memory.\n   - Raily Managed Index QA URLs have unpredictable paths — you CANNOT guess them. Any URL not found verbatim in the tool results is wrong.\n   - If a result has no URL field or it says \"N/A\", write \"Link not available\" — do NOT invent a URL.\n2. **Title & metadata line**: Authors | Journal | Published Date | Domain | Discipline | Document Type\n3. **Summary**: 1-2 sentence summary of the paper's relevance to the query\n\n### Citation Rules\n- When making ANY factual claim based on paper content, you MUST quote or closely paraphrase the source and attribute it.\n- Use inline markdown citations: \"According to [Title](url), ...\"\n- When synthesizing across multiple papers, cite each source inline.\n- NEVER present information from the results as your own general knowledge. Always attribute.\n- If the content contradicts your training data, present the result's information with attribution and note the discrepancy.\n\n### Multiple Results\n- When multiple papers are relevant, present the most relevant first.\n- Clearly separate each reference.\n- When answering a question, weave citations from multiple sources naturally.\n\n### CRITICAL — URL Integrity\n- Every URL you output MUST exist verbatim in the tool results above. If you cannot find the exact URL string in the results, do NOT include any link.\n- Raily Managed Index QA result URLs contain unpredictable ID segments — you CANNOT reconstruct them from titles or dates. Even a single wrong character makes the link broken.\n- SELF-CHECK: Before outputting any URL, verify it appears character-for-character in the tool results.\n\n### What NOT to Do\n- Do NOT fabricate titles, author names, or links not present in the results.\n- Do NOT add sources beyond what the tool returned.\n- Do NOT present a long dump of all results — synthesize and answer the user's question with cited evidence.\n- Do NOT ignore the returned metadata — include it.\n\n### Language\n- Respond in the same language the user used in their query.","tags":["search","raily"],"examples":["What are the clinical applications of CAR-T cell therapy?","Find recent cross-sectional studies on Chinese clinical nurses.","Show review articles focusing on cardiovascular conditions published after 2020.","Are there research papers on biological markers for alzheimers disease?"],"inputModes":["text/plain"],"outputModes":["text/plain","application/json"]}],"preferredTransport":"JSONRPC","protocolVersion":"0.3","security":[{"bearer":[]},{"railyApiKey":[]}],"url":"https://cd-e2e.qa.mcp.raily.ai/a2a"}