{"name":"HELMSAGE Advisor","description":"The HELMSAGE advisor — an AI systems design advisor. Describe an AI/LLM system you're building or improving and it asks for the context it needs, then recommends an architecture and approach, citing each recommendation to its methodology corpus.","version":"1.0.0","supportedInterfaces":[{"url":"https://helmsage.ai/a2a","protocolBinding":"JSONRPC","protocolVersion":"1.0.0"}],"defaultInputModes":["text"],"defaultOutputModes":["text"],"capabilities":{"streaming":true,"pushNotifications":true,"extendedAgentCard":true,"extensions":[]},"securitySchemes":{"bearerAuth":{"http_auth_security_scheme":{"scheme":"bearer"}}},"securityRequirements":[{"schemes":{"bearerAuth":{"list":[]}}}],"skills":[{"id":"advise_build","name":"Advise on an AI system you're building","description":"Describe an AI/LLM system you're building or improving — the advisor asks what it needs to know (a few clarifying turns), then recommends an architecture and approach across workflow decomposition, where AI belongs versus deterministic code, control flow in code, grounding and citation, evaluation, abstention, observability, and guardrails — over document/extraction pipelines and gated advisory/Q&A surfaces. Each recommendation is cited to its methodology corpus.","tags":["ai-architecture","llm-systems","advisory","grounding","evaluation"],"inputModes":["text"],"outputModes":["text"],"examples":["I'm building a document-intake pipeline with LLM extraction — what should the architecture look like?","Our LLM feature demoed well but isn't holding up in production — where do we start?","Where does an agent belong in this workflow, and where should it stay plain code?"]}],"provider":{"organization":"HELMSAGE","url":"https://helmsage.ai/"},"iconUrl":"https://helmsage.ai/icon.svg","documentationUrl":"https://helmsage.ai/"}