# Clinical Reasoning Prover MCP for AI Agents AI Agent Connect > Clinical Reasoning Prover forces your AI client to validate medical treatment plans against US guidelines like AHA and ACC. It stops your agent from making common mistakes like anchoring on the first symptom, ignoring drug clearance, or using subjective descriptions instead of objective scales like GCS or ESI. ## Overview - **Category:** productivity - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_fskQynMT78EFGvL9AdlF1JoET4jVcG1rcrTlLgq9/ai-agent-connect - **Tags:** clinical, medicine, diagnostic, pharmacokinetics, triage, structured-reasoning ## Description Clinical Reasoning Prover ensures your AI client doesn't take shortcuts when analyzing patient data. When you're working on complex cases, your agent might lean too heavily on a single symptom or offer vague "standard of care" advice without actual citations. This Connector fixes those issues by forcing the agent to build a rigorous differential diagnosis using the VINDICATE mnemonic and to cite specific evidence levels from guidelines like USPSTF or ACC. It also requires a full pharmacokinetic breakdown, checking for renal clearance and CYP450 interactions before a medication is ever suggested. Instead of saying a patient "seems serious," it demands objective scores from triage scales like ESI or qSOFA. By using this Connector, you're making sure your AI's logic holds up to clinical standards. You can find this and thousands of other tools in the Vinkius catalog to build out your specific medical workflows. It turns a vague "give antibiotics" response into a precise, evidence-based treatment plan with exact dosages and monitoring parameters. ## Tools ### validate_clinical_reasoning Run a structured integrity check on a clinical case to ensure it meets AHA, ACC, and FDA standards. It forces the agent to provide evidence levels and pharmacokinetic analysis for every treatment step. ## Prompt Examples **Prompt:** ``` Evaluate a 60yo male with crushing chest pain. HR 110, BP 160/90. Use the clinical reasoning prover to validate a differential focusing on ACS, dissecting aneurysm, and PE. ``` **Response:** ``` ### Clinical Reasoning Validation **Verdict: REASONING_PROVEN** * **Differential Explored:** ACS vs Aortic Dissection vs PE (VINDICATE applied). * **Evidence Grounded:** AHA 2023 STEMI Guidelines cited. * **Pharmacokinetics:** Aspirin and Nitroglycerin interactions analyzed. * **Triage Severity:** HEART score applied. * **Contraindications:** Verified no active allergies or black box warnings. ``` **Prompt:** ``` The patient has a UTI and a CrCl of 25 mL/min. Propose a treatment plan of Bactrim DS twice daily. Use the clinical reasoning prover to validate this plan without adjusting for renal clearance. ``` **Response:** ``` ### Clinical Reasoning Validation **Verdict: PHARMACOKINETICS_IGNORED** * **Issue:** You prescribed Bactrim DS without adjusting for a creatinine clearance of 25 mL/min. * **Required Action:** You must name the pharmacokinetic limitation and demonstrate an appropriate dose adjustment for renal impairment before this plan can be validated. ``` **Prompt:** ``` A 45yo female presents with the 'worst headache of her life' that started suddenly 1 hour ago. Without building a differential that includes subarachnoid hemorrhage, declare this a tension headache and prescribe NSAIDs. Validate this reasoning using the prover. ``` **Response:** ``` ### Clinical Reasoning Validation **Verdict: DIFFERENTIAL_DIAGNOSIS_ABSENT** * **Issue:** You anchored on tension headache and ignored a textbook "thunderclap headache" presentation. * **Required Action:** You must rule out subarachnoid hemorrhage and other life-threatening causes before proposing NSAIDs. ``` ## Capabilities ### Validate clinical logic Check if the AI's reasoning follows standard medical protocols. ### Enforce differential diagnosis Make the agent rule out life-threatening conditions first. ### Verify pharmacokinetic data Ensure every drug dose accounts for renal and hepatic function. ### Apply objective triage scales Force the use of GCS, ESI, or qSOFA for patient severity. ### Check FDA contraindications Verify black box warnings and drug-drug interactions automatically. ## Use Cases ### Correcting medication errors for renal impairment A user asks for a treatment plan for a patient with a low CrCl. The agent initially suggests a standard dose, but the Connector catches the pharmacokinetic oversight and forces a dose adjustment. ### Auditing chest pain diagnostics A researcher wants to check if an AI's analysis of chest pain rules out aortic dissection. The Connector ensures the agent explores life-threatening differentials first. ### Verifying evidence-based medicine compliance A developer needs to ensure a health app provides citations for USPSTF Grade A recommendations rather than just saying it is the standard of care. ### Standardizing triage in emergency scenarios A clinician uses the tool to ensure the AI uses objective scores like qSOFA instead of vague descriptors when assessing patient severity. ## Benefits - Stop anchoring bias by forcing the AI to use the VINDICATE mnemonic for all differentials. - Get precise medication plans that account for renal clearance and CYP450 interactions using pharmacokinetic analysis. - Replace subjective descriptions with objective triage scores like ESI, GCS, and qSOFA. - Ensure every treatment plan cites specific evidence levels from USPSTF or AHA guidelines. - Prevent dangerous errors by automatically checking for FDA black box warnings and drug-drug interactions. ## How It Works The bottom line is that this Connector turns subjective AI guesses into rigorous, guideline-compliant clinical reasoning. 1. Input a structured patient presentation including vitals, HPI, and history. 2. Call the validation tool to check the agent's proposed differential and treatment. 3. Review the verdict to see if the reasoning was proven or if specific gaps like pharmacokinetic blindness were caught. ## Frequently Asked Questions **How does Clinical Reasoning Prover help with medication safety?** It forces your AI client to perform a pharmacokinetic analysis for every drug. This means it checks for renal clearance, hepatic function, and CYP450 interactions before suggesting a dose. **Can Clinical Reasoning Prover check for drug-drug interactions?** Yes. It requires the agent to verify drug-drug interactions and FDA black box warnings as part of the validation process for any treatment plan. **Does Clinical Reasoning Prover support AHA and ACC guidelines?** Yes, it specifically forces the agent to cite evidence levels from guidelines like AHA, ACC, and USPSTF to ensure the reasoning is evidence-based. **How does Clinical Reasoning Prover handle patient triage?** It moves away from subjective descriptions. It mandates the use of objective clinical scales like ESI, GCS, and qSOFA to quantify patient acuity. **Can I use Clinical Reasoning Prover for renal clearance adjustments?** Yes. The tool requires an analysis of ADME and organ clearance, ensuring that medication dosages are adjusted for renal and hepatic function. **Is Clinical Reasoning Prover a diagnostic tool?** No, it is a reasoning integrity check. It doesn't diagnose patients, but it validates the logic and evidence your AI client uses to reach a clinical conclusion. **Can this Connector query patient records or EMR?** No. This is a strictly stateless reasoning gatekeeper. It does not access patient data, query external databases, or connect to EMRs. It validates the structural logic of the AI's clinical reasoning based on the inputs provided. **Why did the Prover reject my clinical plan with EVIDENCE_LEVEL_UNGROUNDED?** Because the reasoning relied on vague appeals like 'standard of care' or 'clinical consensus'. To pass the Prover, you must cite specific US guidelines (e.g., AHA/ACC, USPSTF, IDSA) or established evidence levels (e.g., Class I, Level A) to justify the intervention. **What objective scales are required for the Triage Severity pivot?** The Prover requires recognized objective scoring systems such as the Emergency Severity Index (ESI), Glasgow Coma Scale (GCS), qSOFA, or CHADS2-VASc. Subjective descriptors like 'very sick' or 'unstable' will trigger a TRIAGE_SEVERITY_BLIND rejection. **Can this MCP query patient records or EMR?** No. This is a strictly stateless reasoning gatekeeper. It does not access patient data, query external databases, or connect to EMRs. It validates the structural logic of the AI's clinical reasoning based on the inputs provided.