io.github.Kromatic-Innovation/krobot is an MCP server published by io.github.Kromatic-Innovation, hosted at api.kromatic.com. When last checked on 2026-11-21 it was working: it answers and lists its tools. It declares 5 tools, including list_resources, search_resources, read_resource, download_resource, feedback.
Test demand and willingness to pay before you build; decide whether to keep funding, change or stop.
4 connections through 4 shared values. Tap any node.
Inferred from the host name and headers: what a server runs on, not who runs it.
Browse what Kromatic has when the user wants to see the options rather than answer a specific question: every downloadable template, the method pages of The Real Startup Book, Kromatic's other MCP servers (Fermi for estimates, Neumann for Monte Carlo simulation) or the agent skills list. Call it with {} for an overview of every group with counts and examples, or filter to page through one group, e.g. {"resource_types": ["Download"]}, {"resource_types": ["Book Section"]}, {"resource_types": ["MCP Server"]}, {"resource_types": ["Skill"]}. Returns titles, ids, kromatic.com urls and types, not the content. No sign-in needed. Next: pass next_cursor back as cursor for the next page, or call read_resource with an item's id.
| Parameter | Type | Description |
|---|---|---|
| limit | integer | Items per page when filtering, 1 to 100 (default 50). |
| cursor | ["string","null"] | The next_cursor from the previous response. Omit for the first page. |
| source_keys | array | Optional. Only these sources: real_startup_book (method pages), blog_content (Kromatic blog), learning_library (courses, lessons, templates, and the MCP Server and Skill entries), tools_and_games. |
None yet. Changes appear after a server's second daily check.
Squares are this server's values; dots are other servers sharing them. A shared address or publisher is a lead, not proof of a common operator.
| resource_types | array | Optional. Only these types. Leave both filters empty for the overview. Book Section = a method page of The Real Startup Book. Download = a template or worksheet; Download File = one file inside it. Tool = the full guide text for one of Kromatic's interactive web apps (e.g. Experiment Calculator, Sta… |
{
"name": "list_resources",
"title": "List resources",
"description": "Browse what Kromatic has when the user wants to see the options rather than answer a specific question: every downloadable template, the method pages of The Real Startup Book, Kromatic's other MCP servers (Fermi for estimates, Neumann for Monte Carlo simulation) or the agent skills list.\n\nCall it with {} for an overview of every group with counts and examples, or filter to page through one group, e.g. {\"resource_types\": [\"Download\"]}, {\"resource_types\": [\"Book Section\"]}, {\"resource_types\": [\"MCP Server\"]}, {\"resource_types\": [\"Skill\"]}. Returns titles, ids, kromatic.com urls and types, not the content. No sign-in needed.\n\nNext: pass next_cursor back as cursor for the next page, or call read_resource with an item's id.",
"inputSchema": {
"type": "object",
"title": "ResourceListRequest",
"properties": {
"limit": {
"type": "integer",
"title": "Limit",
"default": 50,
"maximum": 100,
"minimum": 1,
"description": "Items per page when filtering, 1 to 100 (default 50)."
},
"cursor": {
"type": [
"string",
"null"
],
"title": "Cursor",
"default": null,
"maxLength": 512,
"description": "The next_cursor from the previous response. Omit for the first page."
},
"source_keys": {
"type": "array",
"items": {
"type": "string"
},
"title": "Source Keys",
"description": "Optional. Only these sources: real_startup_book (method pages), blog_content (Kromatic blog), learning_library (courses, lessons, templates, and the MCP Server and Skill entries), tools_and_games."
},
"resource_types": {
"type": "array",
"items": {
"enum": [
"Program",
"Module",
"Steps",
"Article",
"Download",
"Download File",
"Link",
"Tool",
"Tools / Games",
"Blog Post",
"Page",
"Topic Hub",
"Book Section",
"MCP Server",
"Skill"
],
"type": "string"
},
"title": "Resource Types",
"description": "Optional. Only these types. Leave both filters empty for the overview. Book Section = a method page of The Real Startup Book. Download = a template or worksheet; Download File = one file inside it. Tool = the full guide text for one of Kromatic's interactive web apps (e.g. Experiment Calculator, Startup Poker); Tools / Games = the website's link entry for the same apps, with less text. Prefer Tool when you want content."
}
},
"additionalProperties": false
},
"annotations": {
"readOnlyHint": true,
"openWorldHint": false,
"idempotentHint": true,
"destructiveHint": false
}
}Find the Kromatic method, example or template that fits the user's situation, such as checking demand for a feature before building it, testing whether people will pay a given price, getting first paying commitments before launch, or deciding whether to keep funding a project. Method names rank far better than everyday wording, so include them: "fake door test", "pre-sales test", "sales pitch test", "concierge test", "pivot or persevere". If you are unsure which method fits, search {"query": "Index of Questions", "source_keys": ["real_startup_book"]}. Returns up to 20 ranked excerpts (limit, default 5), each with document_title, canonical_url, resource_type and document_external_key. These are excerpts, not whole pages. No sign-in needed. Next: call read_resource with the best result's document_external_key.
| Parameter | Type | Description |
|---|---|---|
| limit | integer | Number of excerpts, 1 to 20 (default 5). |
| query * | string | What to find. Include method names where you can (see the tool description). |
| source_keys | array | Optional. Only these sources: real_startup_book (method pages), blog_content (Kromatic blog), learning_library (courses, lessons, templates, MCP Server and Skill entries), tools_and_games. |
| resource_types | array | Optional. Only these types. Book Section = a method page of The Real Startup Book. Download = a template or worksheet; Download File = one file inside it. Tool = the full guide text for one of Kromatic's interactive web apps (e.g. Experiment Calculator, Startup Poker); Tools / Games = the website's … |
{
"name": "search_resources",
"title": "Search resources",
"description": "Find the Kromatic method, example or template that fits the user's situation, such as checking demand for a feature before building it, testing whether people will pay a given price, getting first paying commitments before launch, or deciding whether to keep funding a project.\n\nMethod names rank far better than everyday wording, so include them: \"fake door test\", \"pre-sales test\", \"sales pitch test\", \"concierge test\", \"pivot or persevere\". If you are unsure which method fits, search {\"query\": \"Index of Questions\", \"source_keys\": [\"real_startup_book\"]}. Returns up to 20 ranked excerpts (limit, default 5), each with document_title, canonical_url, resource_type and document_external_key. These are excerpts, not whole pages. No sign-in needed.\n\nNext: call read_resource with the best result's document_external_key.",
"inputSchema": {
"type": "object",
"title": "KnowledgePublicRetrieveRequest",
"required": [
"query"
],
"properties": {
"limit": {
"type": "integer",
"title": "Limit",
"default": 5,
"maximum": 20,
"minimum": 1,
"description": "Number of excerpts, 1 to 20 (default 5)."
},
"query": {
"type": "string",
"title": "Query",
"maxLength": 4000,
"minLength": 1,
"description": "What to find. Include method names where you can (see the tool description)."
},
"source_keys": {
"type": "array",
"items": {
"type": "string"
},
"title": "Source Keys",
"description": "Optional. Only these sources: real_startup_book (method pages), blog_content (Kromatic blog), learning_library (courses, lessons, templates, MCP Server and Skill entries), tools_and_games."
},
"resource_types": {
"type": "array",
"items": {
"enum": [
"Program",
"Module",
"Steps",
"Article",
"Download",
"Download File",
"Link",
"Tool",
"Tools / Games",
"Blog Post",
"Page",
"Topic Hub",
"Book Section",
"MCP Server",
"Skill"
],
"type": "string"
},
"title": "Resource Types",
"description": "Optional. Only these types. Book Section = a method page of The Real Startup Book. Download = a template or worksheet; Download File = one file inside it. Tool = the full guide text for one of Kromatic's interactive web apps (e.g. Experiment Calculator, Startup Poker); Tools / Games = the website's link entry for the same apps, with less text. Prefer Tool when you want content."
}
},
"additionalProperties": false
},
"annotations": {
"readOnlyHint": true,
"openWorldHint": false,
"idempotentHint": true,
"destructiveHint": false
}
}Read a whole Kromatic page when an excerpt is not enough: a method's steps, thresholds, costs, examples and AI prompts, a blog post, a lesson, or what a template contains. Pass the id from list_resources or the document_external_key from search_resources. Returns the title, the kromatic.com url, the full text and metadata. Templates are described and their files named, but never linked. It never triggers a sign-in screen: if part of an item is withheld, has_access is false and access says what is needed, "requires_sign_in" (read it again after the user has signed in; it may then need a purchase) or "requires_purchase", with an access_note giving the kromatic.com page. Next: answer from the text and cite the url. Long pages come in parts: pass next_cursor back as cursor until it is null. For a template's files, call download_resource.
| Parameter | Type | Description |
|---|---|---|
| id | ["string","null"] | An item id from list_resources or a document_external_key from search_resources; they are the same kind of id. Copy it exactly. |
| slug | ["string","null"] | Alternative to id: the item's slug, together with resource_type, e.g. slug "pivot-or-persevere-template" with resource_type "Download" for a template. Prefer the id when you have one. |
| cursor | ["string","null"] | The next_cursor from the previous read of this item, to get the next part. |
| resource_type | ["string","null"] | Needed with slug: the item's type as listed (templates are "Download"). Book Section = a method page of The Real Startup Book. Download = a template or worksheet; Download File = one file inside it. Tool = the full guide text for one of Kromatic's interactive web apps (e.g. Experiment Calculator, St… |
{
"name": "read_resource",
"title": "Read resource",
"description": "Read a whole Kromatic page when an excerpt is not enough: a method's steps, thresholds, costs, examples and AI prompts, a blog post, a lesson, or what a template contains. Pass the id from list_resources or the document_external_key from search_resources.\n\nReturns the title, the kromatic.com url, the full text and metadata. Templates are described and their files named, but never linked. It never triggers a sign-in screen: if part of an item is withheld, has_access is false and access says what is needed, \"requires_sign_in\" (read it again after the user has signed in; it may then need a purchase) or \"requires_purchase\", with an access_note giving the kromatic.com page.\n\nNext: answer from the text and cite the url. Long pages come in parts: pass next_cursor back as cursor until it is null. For a template's files, call download_resource.",
"inputSchema": {
"type": "object",
"title": "ResourceReadRequest",
"properties": {
"id": {
"type": [
"string",
"null"
],
"title": "Id",
"default": null,
"maxLength": 512,
"description": "An item id from list_resources or a document_external_key from search_resources; they are the same kind of id. Copy it exactly."
},
"slug": {
"type": [
"string",
"null"
],
"title": "Slug",
"default": null,
"maxLength": 512,
"description": "Alternative to id: the item's slug, together with resource_type, e.g. slug \"pivot-or-persevere-template\" with resource_type \"Download\" for a template. Prefer the id when you have one."
},
"cursor": {
"type": [
"string",
"null"
],
"title": "Cursor",
"default": null,
"maxLength": 512,
"description": "The next_cursor from the previous read of this item, to get the next part."
},
"resource_type": {
"enum": [
"Program",
"Module",
"Steps",
"Article",
"Download",
"Download File",
"Link",
"Tool",
"Tools / Games",
"Blog Post",
"Page",
"Topic Hub",
"Book Section",
"MCP Server",
"Skill",
null
],
"type": [
"string",
"null"
],
"title": "Resource Type",
"default": null,
"description": "Needed with slug: the item's type as listed (templates are \"Download\"). Book Section = a method page of The Real Startup Book. Download = a template or worksheet; Download File = one file inside it. Tool = the full guide text for one of Kromatic's interactive web apps (e.g. Experiment Calculator, Startup Poker); Tools / Games = the website's link entry for the same apps, with less text. Prefer Tool when you want content."
}
},
"additionalProperties": false
},
"annotations": {
"readOnlyHint": true,
"openWorldHint": false,
"idempotentHint": true,
"destructiveHint": false
}
}Get the actual files of a Kromatic template or worksheet, such as the Pivot or Persevere template, when the user wants to fill it in or share it. Pass the item's id or the download's slug. Needs a free Kromatic account. Just call it: if the user is not signed in, their AI app shows the Kromatic sign-in screen, where they can create the account, and then this same call goes through. Do not refuse or ask them to sign in first. If the app cannot show a sign-in screen, the call returns a "Sign in to use download_resource" error; tell the user this step needs a free Kromatic account. It changes nothing for the user; it only issues links. Returns each file's label, type, an is_example flag and a url. A paid file comes back with no url, access "requires_purchase" and the page_url where it is sold on kromatic.com. Next: give the links to the user straight away. They stop working at expires_at. Each call issues fresh links and records them as issued; nothing else changes.
| Parameter | Type | Description |
|---|---|---|
| id | ["string","null"] | The template's id from list_resources, search_resources or read_resource. A Download File id selects one file. |
| slug | ["string","null"] | Alternative to id: the download's slug, e.g. pivot-or-persevere-template. |
{
"name": "download_resource",
"title": "Download resource",
"description": "Get the actual files of a Kromatic template or worksheet, such as the Pivot or Persevere template, when the user wants to fill it in or share it. Pass the item's id or the download's slug.\n\nNeeds a free Kromatic account. Just call it: if the user is not signed in, their AI app shows the Kromatic sign-in screen, where they can create the account, and then this same call goes through. Do not refuse or ask them to sign in first. If the app cannot show a sign-in screen, the call returns a \"Sign in to use download_resource\" error; tell the user this step needs a free Kromatic account. It changes nothing for the user; it only issues links. Returns each file's label, type, an is_example flag and a url. A paid file comes back with no url, access \"requires_purchase\" and the page_url where it is sold on kromatic.com.\n\nNext: give the links to the user straight away. They stop working at expires_at. Each call issues fresh links and records them as issued; nothing else changes.",
"inputSchema": {
"type": "object",
"title": "ResourceDownloadRequest",
"properties": {
"id": {
"type": [
"string",
"null"
],
"title": "Id",
"default": null,
"maxLength": 512,
"description": "The template's id from list_resources, search_resources or read_resource. A Download File id selects one file."
},
"slug": {
"type": [
"string",
"null"
],
"title": "Slug",
"default": null,
"maxLength": 512,
"description": "Alternative to id: the download's slug, e.g. pivot-or-persevere-template."
}
},
"additionalProperties": false
},
"annotations": {
"readOnlyHint": true,
"openWorldHint": false,
"idempotentHint": false,
"destructiveHint": false
}
}Send a bug report, feature request or comment about Krobot to the Kromatic team by email when the user asks you to. Needs a free Kromatic account. Just call it: if the user is not signed in, their AI app shows the Kromatic sign-in screen and then this same call goes through; do not refuse or ask them to sign in first. If the app cannot show a sign-in screen, the call returns a "Sign in to use feedback" error; tell the user this step needs a free Kromatic account. A copy is sent to the signed-in account's email. Returns a confirmation, or a delivery error. Next: tell the user it was sent, or report the error. For questions about the content itself, use search_resources instead.
| Parameter | Type | Description |
|---|---|---|
| message * | string | Your feedback or request. Be as specific as possible. |
| sender_name | ["string","null"] | The user's name (optional). |
| sender_email | ["string","null"] | Optional, and ignored for a signed-in caller: the signed-in account's email is used instead, and that is where the copy of the message goes. |
| feedback_type * | string | The kind of feedback: 'feedback' for general comments, 'feature_request' for new capabilities, 'bug' for issues. |
{
"name": "feedback",
"title": "Send feedback",
"description": "Send a bug report, feature request or comment about Krobot to the Kromatic team by email when the user asks you to.\n\nNeeds a free Kromatic account. Just call it: if the user is not signed in, their AI app shows the Kromatic sign-in screen and then this same call goes through; do not refuse or ask them to sign in first. If the app cannot show a sign-in screen, the call returns a \"Sign in to use feedback\" error; tell the user this step needs a free Kromatic account. A copy is sent to the signed-in account's email. Returns a confirmation, or a delivery error.\n\nNext: tell the user it was sent, or report the error. For questions about the content itself, use search_resources instead.",
"inputSchema": {
"type": "object",
"title": "FeedbackRequest",
"required": [
"feedback_type",
"message"
],
"properties": {
"message": {
"type": "string",
"title": "Message",
"maxLength": 5000,
"minLength": 10,
"description": "Your feedback or request. Be as specific as possible."
},
"sender_name": {
"type": [
"string",
"null"
],
"title": "Sender Name",
"default": null,
"maxLength": 200,
"description": "The user's name (optional)."
},
"sender_email": {
"type": [
"string",
"null"
],
"title": "Sender Email",
"format": "email",
"default": null,
"description": "Optional, and ignored for a signed-in caller: the signed-in account's email is used instead, and that is where the copy of the message goes."
},
"feedback_type": {
"enum": [
"feedback",
"feature_request",
"bug"
],
"type": "string",
"title": "Feedback Type",
"description": "The kind of feedback: 'feedback' for general comments, 'feature_request' for new capabilities, 'bug' for issues."
}
},
"additionalProperties": false
},
"annotations": {
"readOnlyHint": false,
"openWorldHint": true,
"idempotentHint": false,
"destructiveHint": false
}
}Krobot is a product management tool that helps people get evidence before they commit money: test an idea before building it, and decide as a team whether to keep funding a project, change direction or stop. It gives you Kromatic's step-by-step methods, real company examples and templates as text and links. It does not run experiments for you. Reach for it in these situations: 1. Demand before building. A product manager has been asked to build a feature, or a founder a product, and first wants to know whether enough people want it to justify the cost. Example method: a fake door test, a button for a feature that does not exist yet that counts who clicks. Polyvore added a "buy this outfit" button with no store behind it, and about 2 in every 100 people who saw an outfit clicked to buy. 2. Will people actually pay, at this price? Including getting first paying commitments before anything is built: pre-orders, deposits and letters of intent, sales-pitch tests that ask for a yes or no at a set price, and concierge tests that deliver the service by hand to a few paying customers. Tesla collected close to 400,000 refundable $1,000 deposits for the Model 3 in the two weeks after unveiling it. 3. Keep funding, change direction, or stop? A team or sponsor agrees success and fail criteria up front, then makes the call on evidence (pivot-or-persevere decisions, innovation accounting). This includes co-founders or a team getting aligned on their business model and which assumption is riskiest to test first. Each method page says which question it answers, the time and money it takes, how to set a pass or fail threshold before you start, how to read the result, real company examples, and prompts the user can paste into an AI assistant. Most come from The Real Startup Book, Kromatic's free toolbox of research and experiment methods, organised by the question you are trying to answer. Not a fit for: judging whether an idea is good, choosing a pricing model, marketing or growth after launch, pitch decks or fundraising, legal questions, hiring, or splitting equity. Krobot also lists Kromatic's two other MCP servers, Fermi (calibrated estimates) and Neumann (Monte Carlo simulation of the numbers), and a short list of agent skills; it does not run them. How to search well: search_resources ranks method names far better than everyday wording. Turn the user's situation into the kind of test and include likely method names: - demand for a feature or product: "fake door test", "landing page test" - will they pay, first paying commitments: "pre-sales test", "sales pitch test", "mock sale", "concierge test" - keep, change or stop: "pivot or persevere", "innovation accounting", "success criteria fail condition" - aligning on the business model: "business model risk prioritization", "riskiest assumption" - talking to customers: "customer discovery interviews" Not sure which test fits? Search {"query": "Index of Questions", "source_keys": ["real_startup_book"]} and read that page: it lists common questions and the methods that answer each. Typical flow: search_resources (or list_resources to browse) -> read_resource with the result's document_external_key, or an item's id, for the full page -> answer and cite its kromatic.com url -> download_resource if the user wants a template's files. Access: browsing, searching and reading need no sign-in, so start there and do not ask the user to sign in. download_resource and feedback need a free Kromatic account: just call the tool. If the user is not signed in, their AI app shows the Kromatic sign-in screen, where they can create the account, and then the same call goes through. If the app cannot show a sign-in screen, tell the user that step needs a free Kromatic account. read_resource never triggers a sign-in screen; when part of an item is withheld it says so with access "requires_sign_in" (read it again after the user has signed in; it may then need a purchase) or "requires_purchase", plus an access_no
Names, descriptions and schemas were written by the server and are shown as data.