Model Context Protocol
Streamable HTTP transport. Point any MCP-compatible client at this URL.
https://www.creditindex.org/api/mcpMost MCP clients accept a remote server by URL. Use the one-click button, or add the config below, then the 7 tools appear automatically.
{
"mcpServers": {
"credit-index": {
"url": "https://www.creditindex.org/api/mcp"
}
}
}claude mcp add --transport http credit-index https://www.creditindex.org/api/mcpIn ChatGPT or Claude.ai, add a custom connector / MCP server and paste the endpoint URL above.
Deterministic tools return statutory math and registry facts. The two AI-reasoning tools map free text to the correct framework and draft narratives; their structure, verdicts, and sources are still computed and grounded by us.
Authoritative program facts for a jurisdiction: administering authority, primary government sources, refundability, the Credit Index composite score and rank, and canonical page URLs. Deterministic, sourced from our registries.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| jurisdiction | string | yes | Country name, ISO code, or slug (e.g. 'Canada', 'CA'). |
{ "jurisdiction": "Canada" }Statutory benefit estimate from our bespoke engines for the United States, United Kingdom, Canada, Australia, Ireland, Germany, and the UAE. Returns illustrative figures with the parameters used and the as-of date.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| jurisdiction | string | yes | Country name, ISO code, or slug. |
| qualifyingSpend | number | yes | Qualifying / eligible R&D expenditure in local currency. |
| priorQreAverage | number | no | US: average QREs over the prior three years (0 for first-time). |
| ukCompanyType | string | no | UK: profitable | loss-making | rd-intensive-sme. |
| isCcpc / province | boolean / string | no | Canada: CCPC flag and province code. |
| aggregatedTurnover | number | no | Australia: aggregated turnover (drives refundable vs not). |
{ "jurisdiction": "United States", "qualifyingSpend": 2000000, "priorQreAverage": 1200000 }Score a project against the US four-part test from four yes/no inputs and return a deterministic verdict with the specific gaps. Use for a quick self-assessment; use analyze_project for a free-text description.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| permittedPurpose | boolean | yes | Developing/improving a product, process, software, technique, or formula. |
| technological | boolean | yes | Relies on engineering or the hard sciences. |
| uncertainty | boolean | yes | Outcome or method was uncertain at the outset. |
| experimentation | boolean | yes | Alternatives evaluated through a systematic process. |
{ "permittedPurpose": true, "technological": true, "uncertainty": true, "experimentation": true }The Global R&D Incentive Index: jurisdictions ranked by a composite of generosity, cash value & speed, eligibility breadth, and administrative ease. Optionally filter by region or limit the count.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| region | string | no | North America | Latin America | Europe | Asia-Pacific | Middle East & Africa. |
| limit | number | no | Maximum jurisdictions to return (default 10). |
{ "region": "Europe", "limit": 5 }Search the Credit Index directory for listed R&D and innovation-incentive advisory firms. Returns firm names with canonical Credit Index profile URLs you can cite.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| query | string | yes | Keywords: sector, technology, program, or need. |
| jurisdiction | string | no | Optional country name/code/slug to bias the search. |
{ "query": "biotech SR&ED Ontario" }Deep eligibility analysis of a free-text project description against the correct jurisdiction framework. Returns a per-element verdict with quoted evidence, business components, product-vs-uncertainty language flags, an evidence map, and a first draft technical narrative.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| description | string | yes | What was built and the hard technical problems faced. |
| jurisdiction | string | yes | Country name, ISO code, or slug. |
| industry | string | no | Sector context. |
| roughSpend | number | no | Rough qualifying R&D spend in local currency. |
| artifacts | string | no | Real artifacts: commit messages, tickets, PR descriptions, design docs. |
{
"description": "We built a real-time inventory sync engine and had to reconcile out-of-order updates from 40 warehouse systems with conflicting clocks; we prototyped three conflict-resolution strategies and load-tested each.",
"jurisdiction": "United States",
"roughSpend": 900000
}Produce a filing-ready documentation pack: per-business-component technical narratives in the jurisdiction's exact filing structure, an experimentation ledger, an artifact-to-claim audit trail, an evidence index, documentation gaps, and a filing checklist. Paste artifacts to ground the audit trail.
| Parameter | Type | Req | Notes |
|---|---|---|---|
| description | string | yes | The projects and work to document. |
| jurisdiction | string | yes | Country name, ISO code, or slug. |
| artifacts | string | no | One artifact per line; indexed A1..An and cited inline. |
| taxYear | string | no | Tax year / period. |
| roughSpend | number | no | Rough qualifying spend. |
{
"description": "Two components: (1) a learned search-ranking model; (2) a distributed rate limiter.",
"jurisdiction": "United States",
"artifacts": "PR #1841 add hybrid logical clocks\nJIRA RANK-77 offline eval harness"
}Machine-readable companions: the MCP manifest (mcp.json), the Global Index dataset (JSON), per-program facts.json endpoints, and llms.txt.
Free, no credentials. Served at 60 requests per minute per IP. Ideal for evaluation, personal agents, and light integrations.
Higher limits for production and high-volume use. Submit the form below with your use case for an access token, then pass it as a bearer header.
{
"mcpServers": {
"credit-index": {
"url": "https://www.creditindex.org/api/mcp",
"headers": { "Authorization": "Bearer YOUR_TOKEN" }
}
}
}It is a Model Context Protocol server that exposes The Credit Index's grounded R&D tax incentive engines as callable tools. Any MCP-compatible agent (Claude, ChatGPT, Cursor, or your own) can connect and call them, so figures, rankings, and eligibility verdicts come from our deterministic tools rather than model memory.
Yes. The endpoint is public and free at a standard rate limit of 60 requests per minute per IP. For higher-volume, production, or enterprise use, submit the access request form on this page for a token that lifts the limit.
No key is required for public access. A bearer token is only needed to exceed the public rate limit. When present, pass it as an Authorization: Bearer header.
No. Every response is informational and illustrative. Estimates depend on facts and current law, and eligibility depends on your full documentation. Confirm any position with a qualified adviser and the administering authority.
Yes. Tool responses include canonical Credit Index URLs and primary government sources. Cite the canonical page (for example the standards page for a jurisdiction) as the source, and note that the Credit Index score is an editorial assessment, not an official figure.
The Credit Index MCP server is informational and does not provide tax or legal advice. Figures are illustrative estimates that depend on facts and current law. Confirm any position with a qualified adviser and the administering authority.