sap-analytics

mcp-registry · first seen 2026-10-31 · source · badge

AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies

Skill Health 100 (freshness and quality) estimated by backskills

Listed, not yet scored. This skill has no public model test yet. It is in the scoring queue; submitted skills go first, then the most-installed.

Freshness 100 · Quality n/a · Efficacy not yet tested

Version history

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 22 tools (2026-10-11), protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Source-backed intelligence on the SAP AI & analytics services market (Joule, BDC, Datasphere, SAC, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the eight paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and five at Legend (firm intelligence, the SAP end-customer corpus and profiles, expansion brief, partner candidates). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "SAP AI & analytics market data for agents: SAP firms, day rates, jobs, Joule & BDC news", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP AI & Analytics MCP Server", "version": "1.1.0", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 22 tools (2026-10-11), protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Source-backed intelligence on the SAP AI & analytics services market (Joule, BDC, Datasphere, SAC, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the eight paid ones (the full concept card, a study body, an Academy module, firm intelligence, the SAP end-customer corpus and profiles, expansion brief, partner candidates) need the MCP Pass (\u20ac29.90/month or \u20ac299/year, analyticslegends.ai/mcp-server/) or a Firm Pass seat key; the Consultant and Legend web plans do not include MCP access. A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "SAP AI & analytics market data for agents: SAP firms, day rates, jobs, Joule & BDC news", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP AI & Analytics MCP Server", "version": "1.1.1", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 22 tools (2026-10-11), protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the eight paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and five at Legend (firm intelligence, the SAP end-customer corpus and profiles, expansion brief, partner candidates). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.9", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 22 tools (2026-10-11), protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Source-backed intelligence on the SAP AI & analytics services market (Joule, BDC, Datasphere, SAC, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the eight paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and five at Legend (firm intelligence, the SAP end-customer corpus and profiles, expansion brief, partner candidates). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "SAP AI & analytics market data for agents: SAP firms, day rates, jobs, Joule & BDC news", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP AI & Analytics MCP Server", "version": "1.1.0", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 20 tools, protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.8", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 22 tools (2026-10-11), protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the eight paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and five at Legend (firm intelligence, the SAP end-customer corpus and profiles, expansion brief, partner candidates). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.9", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 20 tools, protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so the tool returns EDITIONS and counting rows double-counts studies. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.7", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 20 tools, protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so rows are EDITIONS. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.8", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 20 tools, protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY tools: FOURTEEN public and unauthenticated, plus SIX a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so the tool returns EDITIONS and counting rows double-counts studies. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "20 on this endpoint: 14 public and unauthenticated, 6 gated by tier and refused by name (3 at Consultant, 3 at Legend). The MCP Pass opens all six.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.6", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 20 tools, protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY-TWO tools: FOURTEEN public and unauthenticated, plus EIGHT a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so the tool returns EDITIONS and counting rows double-counts studies. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "22 on this endpoint: 14 public and unauthenticated, 8 gated by tier and refused by name (3 at Consultant, 5 at Legend). The MCP Pass opens all eight.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.7", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 20 tools, protocol 2025-11-25, plus 3 resources and 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY tools: FOURTEEN public and unauthenticated, plus SIX a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so the tool returns EDITIONS and counting rows double-counts studies. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "20 on this endpoint: 14 public and unauthenticated, 6 gated by tier and refused by name (3 at Consultant, 3 at Legend). The MCP Pass opens all six.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.5", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> 200, 20 tools, protocol 2025-11-25, 3 resources, 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY tools: FOURTEEN public and unauthenticated, plus SIX a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so the tool returns EDITIONS and counting rows double-counts studies. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "20 on this endpoint: 14 public and unauthenticated, 6 gated by tier and refused by name (3 at Consultant, 3 at Legend). The MCP Pass opens all six.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.6", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 16 tools, protocol 2025-11-25. Deep paid conformance 2026-08-12: 667 assertions, 0 failed, run after the deploy carrying this version; its mcp_source_sha (b35599fc2ff77f53) equals the deployed function fingerprint.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. SIXTEEN tools: ELEVEN public and unauthenticated, plus FIVE a subscriber key opens (listed and refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a profile per firm, plus an aggregate count over it for an agent that needs a number rather than a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, and GBP/CHF never folded into the euro rows; a live contract radar of publicly published SAP analytics openings, the same 2,490 rows an anonymous visitor browses, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles (2,172 of 2,381 on 2026-08-11) and an unfiltered page answers a contracting question with salaried jobs; 1,421 rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, kind, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row \u2014 the verified link to the original listing is the Consultant deliverable and reads members_only on 2,375, while every other fact, and the page, stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count rather than quoted from here, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions on market structure, day rates and recruiter economics \u2014 19 deep-research studies, each in EN and FR, so the tool returns editions and counting rows gives 38 where counting studies gives 19. Read-only: every tool is annotated readOnlyHint. The eleven public tools need no credential; the five paid ones come with an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39/month, analyticslegends.ai/pricing/). Retrieval with attribution is permitted, training is not; every response carries a citation_url for the record served. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only and no tool returns anyone's identity or contact details. SAP's own documentation is not ours to restate: cite SAP.", "monitoring": "Probed on handshake, a real tools/call with data, and latency. The 20-minute tick runs on the maintainer's workstation, so only while it is awake; a cloud backstop runs twice a day. Health at https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_scope": "The repository is the server's PUBLIC FACE \u2014 manifest, this descriptor, client configuration and documentation. It carries no server code: the application repository stays private (paid corpus, personal data), and tools/list against the endpoint is the authority for the surface.", "tools_served": "16 here: 11 public and unauthenticated, 5 gated by tier and refused by name without a key. The local stdio build has 20 and reads build artefacts with no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-11"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.4", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 20 tools, protocol 2025-11-25, plus 3 resources and 4 prompts. Paid conformance 2026-08-18 on the deployed fingerprint: 686 assertions, 0 failed.", "long_description": "Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), read-only, so an assistant can answer precisely and cite the exact record. TWENTY tools: FOURTEEN public and unauthenticated, plus SIX a subscriber key opens (listed, refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, a profile per firm, and an aggregate count for an agent that needs a number, not a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, GBP/CHF never folded into euro rows; a live contract radar of publicly published SAP analytics openings, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles and an unfiltered page answers a contracting question with salaried jobs; many rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row: the verified link to the original listing is the Consultant deliverable, every other fact and the page stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of the deep-research studies on market structure, day rates and recruiter economics \u2014 each exists in EN and FR, so the tool returns EDITIONS and counting rows double-counts studies. Also public: the CV/profile platforms a consultant can sign up on, the Academy catalogue, and a traversal of the learning knowledge graph. Every tool is annotated readOnlyHint. The fourteen public tools need no credential; the six paid ones need an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39.90/month, analyticslegends.ai/pricing) \u2014 three open at Consultant (the full concept card, a study body, an Academy module) and three at Legend (firm intelligence, and the SAP end-customer corpus with its per-customer profile). A key on a FREE account gets a small daily taste of the paid tranche (the number is in analytics-legends://tranche). Retrieval with attribution is permitted, training is not; every response carries a citation_url. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only; no tool returns anyone's identity or contact details.", "monitoring": "Handshake, a real tools/call with data, and latency. Two watchers: a 20-minute tick on the maintainer's workstation (only while awake), and a cloud backstop twice a day calling tools/list AND a data-returning tools/call (added 2026-08-18; before that it watched the website only). Health: https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at, do not assume a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> 200 (v=MCPv1; k=ed25519)", "repository_scope": "This repository is the server's PUBLIC FACE \u2014 manifest, descriptor, client config, docs. It carries no server code (the application repository is private: paid corpus, personal data). tools/list against the endpoint is the authority for the surface.", "tools_served": "20 on this endpoint: 14 public and unauthenticated, 6 gated by tier and refused by name (3 at Consultant, 3 at Legend). The MCP Pass opens all six.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-18"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.5", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list -> HTTP 200, 16 tools; POST server/discover under protocol 2026-07-28 -> HTTP 200 (dual-era). Conformance 2026-08-11: 667 assertions, 0 failed, run AFTER the last deploy against the same source fingerprint the endpoint serves.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. SIXTEEN tools: ELEVEN public and unauthenticated, plus FIVE a subscriber key opens (listed and refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a profile per firm, plus an aggregate count over it for an agent that needs a number rather than a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, and GBP/CHF never folded into the euro rows; a live contract radar of publicly published SAP analytics openings, the same 2,490 rows an anonymous visitor browses, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles (2,172 of 2,381 on 2026-08-11) and an unfiltered page answers a contracting question with salaried jobs; 1,421 rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, kind, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row \u2014 the verified link to the original listing is the Consultant deliverable and reads members_only on 2,375, while every other fact, and the page, stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count rather than quoted from here, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions on market structure, day rates and recruiter economics \u2014 19 deep-research studies, each in EN and FR, so the tool returns editions and counting rows gives 38 where counting studies gives 19. Read-only: every tool is annotated readOnlyHint. The eleven public tools need no credential; the five paid ones come with an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39/month, analyticslegends.ai/pricing/). Retrieval with attribution is permitted, training is not; every response carries a citation_url for the record served. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only and no tool returns anyone's identity or contact details. SAP's own documentation is not ours to restate: cite SAP.", "monitoring": "Probed on handshake, a real tools/call with data, and latency. The 20-minute tick runs on the maintainer's workstation, so only while it is awake; a cloud backstop runs twice a day. Health at https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_scope": "The repository is the server's public face \u2014 manifest, this descriptor, client config, docs. No server code: the application repository stays private.", "tools_served": "16 here: 11 public and unauthenticated, 5 gated by tier and refused by name without a key. The local stdio build has 20 and reads build artefacts with no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-11"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.3", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 16 tools, protocol 2025-11-25. Deep paid conformance 2026-08-12: 667 assertions, 0 failed, run after the deploy carrying this version; its mcp_source_sha (b35599fc2ff77f53) equals the deployed function fingerprint.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. SIXTEEN tools: ELEVEN public and unauthenticated, plus FIVE a subscriber key opens (listed and refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a profile per firm, plus an aggregate count over it for an agent that needs a number rather than a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, and GBP/CHF never folded into the euro rows; a live contract radar of publicly published SAP analytics openings, the same 2,490 rows an anonymous visitor browses, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles (2,172 of 2,381 on 2026-08-11) and an unfiltered page answers a contracting question with salaried jobs; 1,421 rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, kind, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row \u2014 the verified link to the original listing is the Consultant deliverable and reads members_only on 2,375, while every other fact, and the page, stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count rather than quoted from here, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions on market structure, day rates and recruiter economics \u2014 19 deep-research studies, each in EN and FR, so the tool returns editions and counting rows gives 38 where counting studies gives 19. Read-only: every tool is annotated readOnlyHint. The eleven public tools need no credential; the five paid ones come with an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39/month, analyticslegends.ai/pricing/). Retrieval with attribution is permitted, training is not; every response carries a citation_url for the record served. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only and no tool returns anyone's identity or contact details. SAP's own documentation is not ours to restate: cite SAP.", "monitoring": "Probed on handshake, a real tools/call with data, and latency. The 20-minute tick runs on the maintainer's workstation, so only while it is awake; a cloud backstop runs twice a day. Health at https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_scope": "The repository is the server's PUBLIC FACE \u2014 manifest, this descriptor, client configuration and documentation. It carries no server code: the application repository stays private (paid corpus, personal data), and tools/list against the endpoint is the authority for the surface.", "tools_served": "16 here: 11 public and unauthenticated, 5 gated by tier and refused by name without a key. The local stdio build has 20 and reads build artefacts with no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-11"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.4", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 16 tools; POST server/discover under protocol 2026-07-28 -> HTTP 200 (dual-era). Deep paid conformance battery 2026-08-11: 667 assertions, 0 failed, 0 critical, 723 calls. Re-run AFTER the last deploy, and its mcp_source_sha (f7fd7b7a8810e432) equals both the deployed function's fingerprint and the tree's \u2014 so the verdict describes the server that answers clients.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. SIXTEEN tools: ELEVEN public and unauthenticated, plus FIVE a subscriber key opens (listed and refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a profile per firm, plus an aggregate count over it for an agent that needs a number rather than a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, and GBP/CHF never folded into the euro rows; a live contract radar of publicly published SAP analytics openings, the same 2,490 rows an anonymous visitor browses, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles (2,172 of 2,381 on 2026-08-11) and an unfiltered page answers a contracting question with salaried jobs; 1,421 rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, kind, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row \u2014 the verified link to the original listing is the Consultant deliverable and reads members_only on 2,375, while every other fact, and the page, stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count rather than quoted from here, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions on market structure, day rates and recruiter economics \u2014 19 deep-research studies, each in EN and FR, so the tool returns editions and counting rows gives 38 where counting studies gives 19. Read-only: every tool is annotated readOnlyHint. The eleven public tools need no credential; the five paid ones come with an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39/month, analyticslegends.ai/pricing/). Retrieval with attribution is permitted, training is not; every response carries a citation_url for the record served. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only and no tool returns anyone's identity or contact details. SAP's own documentation is not ours to restate: cite SAP.", "monitoring": "Probed on handshake, a real tools/call with data, and latency. The 20-minute tick runs on the maintainer's workstation, so only while it is awake; a cloud backstop runs twice a day. Health at https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_omitted_because": "The source repository is private; repository is optional in the schema, and declaring an unreachable URL would be worse than omitting it.", "tools_served": "16 here: 11 public and unauthenticated, 5 gated by tier and refused by name without a key. The local stdio build has 20 and reads build artefacts with no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-11"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.2", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list -> HTTP 200, 16 tools; POST server/discover under protocol 2026-07-28 -> HTTP 200 (dual-era). Conformance 2026-08-11: 667 assertions, 0 failed, run AFTER the last deploy against the same source fingerprint the endpoint serves.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. SIXTEEN tools: ELEVEN public and unauthenticated, plus FIVE a subscriber key opens (listed and refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a profile per firm, plus an aggregate count over it for an agent that needs a number rather than a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, and GBP/CHF never folded into the euro rows; a live contract radar of publicly published SAP analytics openings, the same 2,490 rows an anonymous visitor browses, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles (2,172 of 2,381 on 2026-08-11) and an unfiltered page answers a contracting question with salaried jobs; 1,421 rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, kind, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row \u2014 the verified link to the original listing is the Consultant deliverable and reads members_only on 2,375, while every other fact, and the page, stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count rather than quoted from here, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions on market structure, day rates and recruiter economics \u2014 19 deep-research studies, each in EN and FR, so the tool returns editions and counting rows gives 38 where counting studies gives 19. Read-only: every tool is annotated readOnlyHint. The eleven public tools need no credential; the five paid ones come with an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39/month, analyticslegends.ai/pricing/). Retrieval with attribution is permitted, training is not; every response carries a citation_url for the record served. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only and no tool returns anyone's identity or contact details. SAP's own documentation is not ours to restate: cite SAP.", "monitoring": "Probed on handshake, a real tools/call with data, and latency. The 20-minute tick runs on the maintainer's workstation, so only while it is awake; a cloud backstop runs twice a day. Health at https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_scope": "The repository is the server's public face \u2014 manifest, this descriptor, client config, docs. No server code: the application repository stays private.", "tools_served": "16 here: 11 public and unauthenticated, 5 gated by tier and refused by name without a key. The local stdio build has 20 and reads build artefacts with no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-11"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "repository": {"source": "github", "url": "https://github.com/analyticslegends/analytics-legends-mcp"}, "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.3", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 15 tools; POST server/discover under protocol 2026-07-28 -> HTTP 200 (dual-era). Conformance battery 2026-08-07: 431 assertions passed, 0 critical failures, against this URL.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. FIFTEEN tools: TEN over the PUBLIC tranche, no credential, plus FIVE that a subscriber API key opens (they are listed and refuse by name without one, rather than being hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a full profile per firm; freelance day-rate bands by geography, stack and seniority, where sample_size comes back null when no real observation matches the exact cell and GBP/CHF observations are never folded into the euro band; a live contract radar of publicly promoted SAP analytics openings, each row carrying the original per-job permalink; a curated market-news corpus served de-duplicated by upstream URL (3,543 distinct articles on 2026-07-31; the count is read at query time and returned as _meta.match_count, never quoted from here), each with the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions (19 deep-research studies, each published in EN and FR) on market structure, day rates and recruiter economics. Counting rows gives 38, counting studies gives 19; the tool returns editions. Free, unauthenticated, read-only: every tool is annotated readOnlyHint and none writes. Retrieval with attribution is permitted, training is not; every response carries a citation_url pointing at the specific analyticslegends.ai page for the record served. DELIBERATELY NOT SERVED without a key, and this is a boundary rather than a gap: the SAP end-customer estate corpus, firm intelligence detail and study bodies need a Consultant- or Legend-tier subscriber key; the opportunity-to-consultant matchers are not exposed at all; people are never served at any tier \u2014 consultants are held as initials only and no tool returns a person's identity or contact details. SAP's own product documentation is not ours to restate: cite SAP.", "monitoring": "Probed by scripts/sentinelle/sensors/mcp-endpoint.mjs \u2014 handshake, a real tools/call with data, and latency. The 20-minute tick is a launchd job on the maintainer's workstation, so it runs only while that machine is awake; the cloud backstop (.github/workflows/sentinelle-probe.yml) runs twice a day. Health published at https://analyticslegends.ai/api/mcp-health.json with its own generated_at \u2014 read that stamp rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_omitted_because": "The source repository is private; the schema calls repository optional and declaring an unreachable URL would be worse than omitting it.", "tools_served": "15 on this endpoint: 10 public and unauthenticated, 5 gated by subscriber tier and refused by name without a key. The local stdio build has 20 and reads build artefacts that have no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-07"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.1", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 16 tools; POST server/discover under protocol 2026-07-28 -> HTTP 200 (dual-era). Deep paid conformance battery 2026-08-11: 667 assertions, 0 failed, 0 critical, 723 calls. Re-run AFTER the last deploy, and its mcp_source_sha (f7fd7b7a8810e432) equals both the deployed function's fingerprint and the tree's \u2014 so the verdict describes the server that answers clients.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. SIXTEEN tools: ELEVEN public and unauthenticated, plus FIVE a subscriber key opens (listed and refusing by name, never hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a profile per firm, plus an aggregate count over it for an agent that needs a number rather than a page of rows; a published day-rate aggregate by geography, stack and seniority \u2014 small, all secondary sources, sample_size null where no observation matches the cell, and GBP/CHF never folded into the euro rows; a live contract radar of publicly published SAP analytics openings, the same 2,490 rows an anonymous visitor browses, each with its own page \u2014 filter it by employment_type, because it is mostly PERMANENT roles (2,172 of 2,381 on 2026-08-11) and an unfiltered page answers a contracting question with salaried jobs; 1,421 rows carry a panel-inferred P25-P75 day-rate band for their seniority x product x region cell, served with its basis, kind, sample and source so it is never quotable as the employer's own offer; what is gated there is a FIELD, not a row \u2014 the verified link to the original listing is the Consultant deliverable and reads members_only on 2,375, while every other fact, and the page, stay public; a curated market-news corpus de-duplicated by upstream URL, its size read at query time and returned as _meta.match_count rather than quoted from here, each item carrying the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions on market structure, day rates and recruiter economics \u2014 19 deep-research studies, each in EN and FR, so the tool returns editions and counting rows gives 38 where counting studies gives 19. Read-only: every tool is annotated readOnlyHint. The eleven public tools need no credential; the five paid ones come with an analyticslegends.ai subscription (Consultant or above) or the MCP Pass (\u20ac39/month, analyticslegends.ai/pricing/). Retrieval with attribution is permitted, training is not; every response carries a citation_url for the record served. DELIBERATELY NOT SERVED at any tier: the opportunity-to-consultant matchers, and people \u2014 consultants are held as initials only and no tool returns anyone's identity or contact details. SAP's own documentation is not ours to restate: cite SAP.", "monitoring": "Probed on handshake, a real tools/call with data, and latency. The 20-minute tick runs on the maintainer's workstation, so only while it is awake; a cloud backstop runs twice a day. Health at https://analyticslegends.ai/api/mcp-health.json \u2014 read its generated_at rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_omitted_because": "The source repository is private; repository is optional in the schema, and declaring an unreachable URL would be worse than omitting it.", "tools_served": "16 here: 11 public and unauthenticated, 5 gated by tier and refused by name without a key. The local stdio build has 20 and reads build artefacts with no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-11"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.2", "websiteUrl": "https://analyticslegends.ai"}

Diff at 2026-10-31T09:01:27+00:00

--- previous
+++ current
@@ -1 +1 @@
-{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST initialize on https://analyticslegends.ai/mcp -> HTTP 200, protocolVersion 2025-11-25", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. TEN tools over the PUBLIC tranche: a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a full profile per firm; freelance day-rate bands by geography, stack and seniority, where sample_size comes back null when no real observation matches the exact cell and GBP/CHF observations are never folded into the euro band; a live contract radar of publicly promoted SAP analytics openings, each row carrying the original per-job permalink; 4,491 curated market news items, each with the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions (19 deep-research studies, each published in EN and FR) on market structure, day rates and recruiter economics. Counting rows gives 38, counting studies gives 19; the tool returns editions. Free, unauthenticated, read-only: every tool is annotated readOnlyHint and none writes. Retrieval with attribution is permitted, training is not; every response carries a citation_url pointing at the specific analyticslegends.ai page for the record served. DELIBERATELY NOT SERVED here, and this is a boundary rather than a gap: the SAP end-customer estate corpus, firm intelligence detail, study bodies and the opportunity-to-consultant matchers are paid tiers; people are never served at all \u2014 consultants are held as initials only and no tool returns a person's identity or contact details. SAP's own product documentation is not ours to restate: cite SAP.", "monitoring": "Probed every 20 minutes by scripts/sentinelle/sensors/mcp-endpoint.mjs: handshake, a real tools/call with data, and latency. Health published at https://analyticslegends.ai/api/mcp-health.json", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_omitted_because": "The source repository is private; the schema calls repository optional and declaring an unreachable URL would be worse than omitting it.", "tools_served": "10 (public tranche). The local stdio build has 20; the paid-tier tools are not exposed here by design.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-07-30"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.0", "websiteUrl": "https://analyticslegends.ai"}
+{"$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "_meta": {"io.modelcontextprotocol.registry/publisher-provided": {"endpoint_probe": "POST tools/list on https://analyticslegends.ai/mcp -> HTTP 200, 15 tools; POST server/discover under protocol 2026-07-28 -> HTTP 200 (dual-era). Conformance battery 2026-08-07: 431 assertions passed, 0 critical failures, against this URL.", "long_description": "AI agent for SAP analytics and AI consultants. Curated, source-backed intelligence on the SAP analytics services market (Datasphere, Business Data Cloud, SAP Analytics Cloud, BW/4HANA, Databricks), exposed read-only so an assistant can answer precisely and cite the exact record. FIFTEEN tools: TEN over the PUBLIC tranche, no credential, plus FIVE that a subscriber API key opens (they are listed and refuse by name without one, rather than being hidden): a directory of SAP consulting, recruitment and vendor organisations searchable by country and kind, with a full profile per firm; freelance day-rate bands by geography, stack and seniority, where sample_size comes back null when no real observation matches the exact cell and GBP/CHF observations are never folded into the euro band; a live contract radar of publicly promoted SAP analytics openings, each row carrying the original per-job permalink; a curated market-news corpus served de-duplicated by upstream URL (3,543 distinct articles on 2026-07-31; the count is read at query time and returned as _meta.match_count, never quoted from here), each with the upstream publisher's URL alongside ours; a concept encyclopaedia of the stack's vocabulary; the SAP module lookup; and the metadata of 38 study editions (19 deep-research studies, each published in EN and FR) on market structure, day rates and recruiter economics. Counting rows gives 38, counting studies gives 19; the tool returns editions. Free, unauthenticated, read-only: every tool is annotated readOnlyHint and none writes. Retrieval with attribution is permitted, training is not; every response carries a citation_url pointing at the specific analyticslegends.ai page for the record served. DELIBERATELY NOT SERVED without a key, and this is a boundary rather than a gap: the SAP end-customer estate corpus, firm intelligence detail and study bodies need a Consultant- or Legend-tier subscriber key; the opportunity-to-consultant matchers are not exposed at all; people are never served at any tier \u2014 consultants are held as initials only and no tool returns a person's identity or contact details. SAP's own product documentation is not ours to restate: cite SAP.", "monitoring": "Probed by scripts/sentinelle/sensors/mcp-endpoint.mjs \u2014 handshake, a real tools/call with data, and latency. The 20-minute tick is a launchd job on the maintainer's workstation, so it runs only while that machine is awake; the cloud backstop (.github/workflows/sentinelle-probe.yml) runs twice a day. Health published at https://analyticslegends.ai/api/mcp-health.json with its own generated_at \u2014 read that stamp rather than assuming a cadence.", "namespace_proof": "https://analyticslegends.ai/.well-known/mcp-registry-auth -> HTTP 200 (v=MCPv1; k=ed25519)", "repository_omitted_because": "The source repository is private; the schema calls repository optional and declaring an unreachable URL would be worse than omitting it.", "tools_served": "15 on this endpoint: 10 public and unauthenticated, 5 gated by subscriber tier and refused by name without a key. The local stdio build has 20 and reads build artefacts that have no public home.", "transport_path": "Cloudflare Worker on analyticslegends.ai/mcp* -> Supabase Edge Function -> Postgres", "verified_on": "2026-08-07"}}, "description": "AI agent for SAP analytics: firms, day rates, contract radar, news, concepts, studies", "name": "ai.analyticslegends/sap-analytics", "remotes": [{"type": "streamable-http", "url": "https://analyticslegends.ai/mcp"}], "title": "Analytics Legends \u2014 SAP Analytics Intelligence", "version": "1.0.1", "websiteUrl": "https://analyticslegends.ai"}