# Archimedes First Principles Prover MCP > Archimedes First Principles Prover forces your AI agent to prove its reasoning from foundational axioms, rather than relying on industry analogies or obvious assumptions. This MCP systematically decomposes complex problems into irreducible parts and identifies true failure boundaries, ensuring every conclusion is logically derived. ## Overview - **Category:** architecture - **Price:** Free - **Tags:** first-principles, axiom-reasoning, decomposition, proof-derivation, boundary-testing, leverage-point, archimedes, method-of-exhaustion ## Description When you need an answer that holds up under intense scrutiny, this connector provides the rigor of a mathematical proof. It stops your agent from simply repeating what 'the industry leader does' or asserting a conclusion because it feels obvious. Instead, it forces five distinct analytical pivots: first, defining foundational claims and questioning their source; second, breaking down monolithic problems into interacting components; third, deriving conclusions through an explicit logical chain; fourth, testing the system's limits to find where it fails; and finally, pinpointing the single variable that creates disproportionate output. By running these checks through Vinkius, you ensure your AI agent’s strategy isn't built on pattern matching or analogy—it's built on demonstrable truth. ## Tools ### validate_archimedes_first_principles Runs a deep validation check, forcing your agent to prove its reasoning by examining assumptions, decomposing problems, deriving proofs, testing boundaries, and finding leverage points. ## Prompt Examples **Prompt:** ``` We should adopt a decentralized structure because the top 3 companies in our industry all use decentralized teams. Obviously this is the right approach. ``` **Response:** ``` AXIOM_UNEXAMINED — 'The top companies use decentralized teams' is an analogy, not an axiom. Archimedes derived pi from polygon geometry — he did not copy the Egyptians. At what actual team size does YOUR centralized structure fail? If your organization has 40 people and coordination works, 'we need independent units' is false. 'Obviously' is the word people use when they cannot prove something. State the foundational claim: what constraint makes the current structure impossible at YOUR scale? ``` **Prompt:** ``` Our conversion rate is 2.1% vs. industry average of 3.5%. We should optimize the checkout flow like the leading stores do. ``` **Response:** ``` DECOMPOSITION_ABSENT — Axioms partially pass: 2.1% conversion is measured (good), 3.5% industry average is sourced (good). But 'like leading stores' is analogy contamination. More critically: decomposition is missing. 'Optimize the checkout' treats the funnel as monolithic. WHERE do users drop? Is it the address form (68% of Baymard's data)? Payment step? Shipping cost reveal? Decompose: cart → address → shipping → payment → confirmation. Which step has the highest drop-off? Without decomposition, you optimize the wrong step. Archimedes did not say 'the curve is complex' — he decomposed it into segments. ``` **Prompt:** ``` Our quality inspection takes 340 minutes per batch. We need to restructure the entire organization to fix this bottleneck. ``` **Response:** ``` PROOF_MISSING — 'Restructure the entire organization' is not derived from the axiom. The axiom is measured (340 min — good), but the conclusion jumps to restructuring without proof. WHY does 340 min mean restructuring? Decompose first: which step in the inspection takes the most time? If it is cross-referencing 3 data sources manually, a standardized template addresses the source — no restructuring needed. Archimedes: 'We must suppose that... therefore it follows that...' Your 'therefore' is missing. What logical chain connects '340 minutes' to 'restructure everything'? ``` ## Capabilities ### Question Foundational Axioms It forces the analysis to state every core assumption and identify its source, preventing reliance on mere industry parallels. ### Decompose Complex Problems The system breaks down a large problem into smaller, independent components, mapping how they interact with one another. ### Derive Logical Proofs It requires the agent to show every step in the conclusion's logical path from the initial axioms. ### Test Operational Boundaries The analysis identifies both minimum and maximum input limits, documenting where the proposed solution starts working and, crucially, where it breaks down. ### Identify Leverage Points It finds the single variable or constraint where a small change in input yields a massive, disproportionate output. ## Use Cases ### Evaluating a new market entry strategy A firm proposes entering Market Z because 'all competitors are doing it.' The agent runs the MCP, which immediately flags this as an unexamined axiom and demands proof of failure points for the current structure at the proposed scale. ### Redesigning a core engineering process An engineer suggests rewriting the entire data pipeline because 'it's inefficient.' The MCP forces decomposition, revealing that the inefficiency is localized to one specific API call, not the whole system, saving months of rework. ### Validating a complex business model The team claims their revenue will scale infinitely. Running the MCP tests boundaries and outputs a warning: 'Unbounded claim,' forcing them to define resource saturation limits and cost ceilings. ## Benefits - Stops reasoning built on analogy. The tool requires the agent to state every foundational claim and source it, eliminating 'Company X did Y' assumptions. - Forces true problem decomposition. You move past treating a system as one big blob by breaking it into its irreducible components and mapping their interactions. - Eliminates guesswork. Instead of asserting conclusions with 'obviously,' the MCP demands a step-by-step, provable logical chain from axioms to outcome. - Finds failure modes. It mandates testing boundaries (min/max input) so you don't launch a product that only works under ideal, unrealistic conditions. - Pinpoints maximum impact. The analysis isolates the single variable—the fulcrum—that provides disproportionate return on effort. ## How It Works The bottom line is, you get an auditable, mathematically rigorous framework for evaluating high-stakes decisions. 1. Input the strategic problem and initial proposed solution to your AI agent. 2. The MCP runs the reasoning through five structured validation pivots: axiom examination, decomposition, proof derivation, boundary testing, and leverage identification. 3. You receive a detailed verdict matrix that flags any structural gap in the reasoning—pointing out if assumptions are unproven or if limits were ignored. ## Frequently Asked Questions **How is this different from the Elon Musk Physics Prover?** Elon Musk Physics Prover forces the 5-Step Starbase Algorithm: question, delete, simplify, accelerate, automate. It is about operational engineering — cutting bloat. Archimedes First Principles Prover forces axiom-based reasoning: state axioms, decompose, prove, test boundaries, find leverage. It is about analytical rigor — proving your logic before building. Musk asks 'should this exist?' Archimedes asks 'is this actually true?' **What counts as a valid axiom?** An axiom is a foundational claim your reasoning depends on, with an explicit source: measurement ('our average processing time is 340 minutes — measured last Tuesday'), physics ('material strength decreases by 15% per 10°C above threshold'), economics ('our acquisition cost exceeds lifetime value at current pricing'), or stated assumption ('we assume retention stays at 14 months'). 'Organization X does Y' is analogy. 'Obviously' is assertion. Neither is an axiom. **Can I use this for business strategy, not just engineering?** Yes. First-principles reasoning applies wherever analogical reasoning misleads. 'We should use freemium because the market leader does' is an analogy. The axiom is: at what conversion rate does freemium generate more lifetime value than paid-only? Decomposition: acquisition, activation, retention, monetization — which component is the actual bottleneck? Proof: if conversion is 3% and free-tier cost is $X/user, then... Boundary: at what scale does free-tier cost exceed premium revenue? Leverage: which single metric, if improved 10%, changes the business? **What kind of data should I feed into `validate_archimedes_first_principles`?** The tool accepts complex scenario descriptions and measurable metrics. Focus on describing a single, high-stakes problem; the quality of your input determines the depth of the analysis. **If I get an error from `validate_archimedes_first_principles`, what does that signify?** An error means your initial reasoning contains a structural flaw. The output will point directly to which principle failed, like AXIOM_UNEXAMINED or DECOMPOSITION_ABSENT. **Do I need specific credentials for `validate_archimedes_first_principles`?** No; you only connect your agent via Vinkius. This MCP is designed to integrate immediately with any compatible client without requiring new keys or setup. **What are the limitations or rate limits when using `validate_archimedes_first_principles`?** There are no usage restrictions on our end. Vinkius manages all necessary API throttling, allowing you to run deep analyses as frequently as needed. **How do I interpret the output of the 'Verdict Matrix' from `validate_archimedes_first_principles`?** The matrix flags exactly which foundational steps were skipped. If all pivots pass, you receive PRINCIPLES_PROVEN, confirming your logic is robust and fully tested.