# ComputeSage machine discovery > Evidence-led private-AI deployment intelligence with a stateless MCP v2 server, bounded public evidence, and accountless x402-paid planning routes. status: available service: ComputeSage / StackBench public agent documentation build_commit: e357cd4 mcp_endpoint: https://mcp.computesage.com/mcp pricing_page: https://mcp.computesage.com/pricing api_docs: https://mcp.computesage.com/api llms_full: https://mcp.computesage.com/llms-full.txt http_base_url: https://mcp.computesage.com network: eip155:8453 asset: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913 currency: USDC (6-decimal atomic units), display_currency: USD payment: x402 v2 exact EVM, per_call; payment is not evidence or deployment proof. availability: public production MCP origin on this host exposes /mcp, /llms.txt, /llms-full.txt, /pricing, /api, and /agent/v1/*; production hosting is distinct from the protected beta hostname; Base mainnet settlement is claimed only by the verified release proof. access_boundary: this public MCP origin does not require Cloudflare Access; beta.computesage.com remains behind the computesage-beta-access / beta-invitees boundary. funded_proof: verified production Base mainnet x402 release evidence includes settled launch-config and evidence-export direct HTTP calls totaling $0.60 with canonical USDC Transfer and ledger correlation; funded MPP mainnet remains unclaimed. mcp_config_json: {"mcpServers":{"computesage":{"url":"https://mcp.computesage.com/mcp"}}} receipt.mcp: result.structuredContent.payment_receipt receipt.http: top-level payment_receipt payment_header.challenge: PAYMENT-REQUIRED payment_header.signature: PAYMENT-SIGNATURE payment_header.response: PAYMENT-RESPONSE payment_header.reconciliation: X-StackBench-Ledger-Reconciliation discovery_side_effects: no x402 verification/settlement, ledger reserve/complete/fail, decision call, export query, persistence, or credential issuance. catalog_json: {"status":"available","mcp_endpoint":"https://mcp.computesage.com/mcp","http_base_url":"https://mcp.computesage.com","tools":[{"name":"search_evidence","description":"Free search of bounded public StackBench benchmark and failure evidence; use this for evidence-only requests."},{"name":"check_deployment_fit","description":"Paid USDC Base mainnet check against existing StackBench can-it-run evidence."},{"name":"predict_performance","description":"Paid configuration performance prediction from existing StackBench evidence; returns measured-supported speed, memory, and power metrics with confidence, missing metrics, evidence, and caveats."},{"name":"recommend_deployment","description":"Paid deployment recommendation for what to buy or run: one supported configuration plus bounded alternatives, performance, confidence, evidence, caveats, and economics only when supported."},{"name":"generate_launch_config","description":"Paid deterministic vLLM or llama.cpp launch guidance derived from an exact StackBench recommendation; commands are copyable templates and do not claim a launch occurred."}],"policies":[{"resource_kind":"mcp_tool","resource_name":"search_evidence","amount_atomic":"0","price_usd":"0.00","currency":"USDC","display_currency":"USD","network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"},{"resource_kind":"mcp_tool","resource_name":"check_deployment_fit","amount_atomic":"10000","price_usd":"0.01","currency":"USDC","display_currency":"USD","network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"},{"resource_kind":"mcp_tool","resource_name":"predict_performance","amount_atomic":"50000","price_usd":"0.05","currency":"USDC","display_currency":"USD","network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"},{"resource_kind":"mcp_tool","resource_name":"recommend_deployment","amount_atomic":"250000","price_usd":"0.25","currency":"USDC","display_currency":"USD","network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"},{"resource_kind":"mcp_tool","resource_name":"generate_launch_config","amount_atomic":"500000","price_usd":"0.50","currency":"USDC","display_currency":"USD","network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"},{"resource_kind":"http_route","resource_name":"evidence_export","amount_atomic":"100000","price_usd":"0.10","currency":"USDC","display_currency":"USD","network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"}],"http_routes":[{"method":"POST","path":"/agent/v1/check-fit","resource_name":"check_deployment_fit","price_usd":"0.01"},{"method":"POST","path":"/agent/v1/predict","resource_name":"predict_performance","price_usd":"0.05"},{"method":"POST","path":"/agent/v1/recommend","resource_name":"recommend_deployment","price_usd":"0.25"},{"method":"POST","path":"/agent/v1/launch-config","resource_name":"generate_launch_config","price_usd":"0.50"},{"method":"GET","path":"/agent/v1/evidence/export","resource_name":"evidence_export","price_usd":"0.10"}],"network":"eip155:8453","asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913","payment":"x402 v2 exact EVM per_call; payment receipt is returned only after successful settlement","receipts":{"mcp":"result.structuredContent.payment_receipt","http":"top-level payment_receipt","challenge":"PAYMENT-REQUIRED","signature":"PAYMENT-SIGNATURE","response":"PAYMENT-RESPONSE","reconciliation":"X-StackBench-Ledger-Reconciliation"}} ## Canonical resources - [MCP endpoint](https://mcp.computesage.com/mcp): Streamable HTTP; use tools/list as the authoritative runtime primitive catalog. - [MCP server card](https://mcp.computesage.com/mcp/server-card): Draft MCP server-card discovery document. - [AI catalog](https://mcp.computesage.com/.well-known/ai-catalog.json): Domain-level pointer to the MCP server card. - [OpenAPI](https://mcp.computesage.com/openapi.json): Canonical public HTTP API schema. - [Pricing](https://mcp.computesage.com/pricing): Active fail-closed policy projection. - [Sitemap](https://mcp.computesage.com/sitemap.xml): Canonical public HTML routes. ## Released MCP tools tool.search_evidence.description: Free search of bounded public StackBench benchmark and failure evidence; use this for evidence-only requests. tool.check_deployment_fit.description: Paid USDC Base mainnet check against existing StackBench can-it-run evidence. tool.predict_performance.description: Paid configuration performance prediction from existing StackBench evidence; returns measured-supported speed, memory, and power metrics with confidence, missing metrics, evidence, and caveats. tool.recommend_deployment.description: Paid deployment recommendation for what to buy or run: one supported configuration plus bounded alternatives, performance, confidence, evidence, caveats, and economics only when supported. tool.generate_launch_config.description: Paid deterministic vLLM or llama.cpp launch guidance derived from an exact StackBench recommendation; commands are copyable templates and do not claim a launch occurred. tool.search_evidence.price_usd: $0.00 tool.check_deployment_fit.price_usd: $0.01 tool.predict_performance.price_usd: $0.05 tool.recommend_deployment.price_usd: $0.25 tool.generate_launch_config.price_usd: $0.50 ## Direct HTTP routes http.POST https://mcp.computesage.com/agent/v1/check-fit: $0.01 per call; x402 response headers and receipt are documented at https://mcp.computesage.com/api http.POST https://mcp.computesage.com/agent/v1/predict: $0.05 per call; x402 response headers and receipt are documented at https://mcp.computesage.com/api http.POST https://mcp.computesage.com/agent/v1/recommend: $0.25 per call; x402 response headers and receipt are documented at https://mcp.computesage.com/api http.POST https://mcp.computesage.com/agent/v1/launch-config: $0.50 per call; x402 response headers and receipt are documented at https://mcp.computesage.com/api http.GET https://mcp.computesage.com/agent/v1/evidence/export: $0.10 per call; x402 response headers and receipt are documented at https://mcp.computesage.com/api http.evidence_export.rights: bounded rights-eligible export only; no raw or private corpus is included in discovery documents.