PreVibe
Xác thực ý tưởng SaaS của bạn bằng nghiên cứu thị trường trước khi xây dựng nó.
Tài liệu
PreVibe MCP Connector
Connect your favorite AI app to your PreVibe account via the Model Context Protocol (MCP). Just add the connector and sign in with Google - no API keys to copy.
Connected apps
AI apps you have connected to your PreVibe account. Revoke access at any time.
Sign in to connect an AI app and manage its access.
Installation Guide
Connect to PreVibe via the Model Context Protocol (MCP). You sign in with Google - there are no API keys to copy.
Server URL
https://previbe.io/api/mcp
Available Commands
Shortcuts your AI app can run against PreVibe. Where they live depends on the app: in Claude Desktop, click + in the message box, choose Connectors, then Add from PreVibe. In Claude Code they appear as /previbe:validate-idea. You can also ask in plain language - say "with PreVibe" to be sure the request runs a real report rather than your assistant's own guess. Or let PreVibe offer, since your assistant knows when to suggest a validation. Reconnect the server if a new command does not show up yet.
/validate-idea
Since 3.3.0
Run grounded research on an idea and get back a GO / PIVOT / NO-GO verdict, a 0-100 viability score, key findings and top bets.
Parameters:
• idea (string, required): The idea to validate - a one-liner or full notes, up to 4,000 characters
Researches
create_research
1 credit
Generate a market research report from a business idea. Accepts anything from a one-liner to full multiline notes. Returns a compact summary - verdict badge, score, one-paragraph summary, key findings, and top bets - plus the research ID; fetch the full report on demand with get_research.
Parameters:
• idea (string, required): Business idea or topic (up to 4,000 characters)
list_researches
Read-onlyFree
List all your researches with pagination support.
Parameters:
• limit (number, optional): Results per page (default: 20, max: 100)
• offset (number, optional): Pagination offset (default: 0)
get_research
Read-onlyFree
Get a specific research by ID, including the full report.
Parameters:
• id (string, required): Research UUID
retry_research
Free
Retry a research whose generation did not complete (failed, or stuck without a result). Free - the credit was already charged when it was created. Returns the same compact summary as create_research.
Parameters:
• id (string, required): Research UUID
delete_research
DestructiveFree
Delete one of your researches by ID. Any deep dives linked to it are removed with it. This cannot be undone.
Parameters:
• id (string, required): Research UUID
Deep Dives
create_deep_dive
1 credit
Generate an in-depth framework analysis (jobs-to-be-done, positioning, MVP scope, Kano, pricing, 90-day plan, risks) for a completed research. Seeded strictly from that research - no new market data. Returns the full analysis.
Parameters:
• researchId (string, required): UUID of the completed research to deep-dive
• target (string, optional): What to analyze: 'idea' (your own idea) or a top bet ('bet-1'/'bet-2'/'bet-3'). Omit for the verdict-resolved default.
list_deep_dives
Read-onlyFree
List all deep dives generated for a specific research.
Parameters:
• researchId (string, required): Research UUID
get_deep_dive
Read-onlyFree
Get a specific deep dive by ID, including its full analysis.
Parameters:
• id (string, required): Deep dive UUID
retry_deep_dive
Free
Retry a deep dive whose generation did not complete. Free - the credit was already charged when it was created. Returns the full regenerated analysis.
Parameters:
• id (string, required): Deep dive UUID
delete_deep_dive
DestructiveFree
Delete a deep dive by ID. This cannot be undone.
Parameters:
• id (string, required): Deep dive UUID
Brand Names
create_brand_names
1 credit
Generate a set of brandable names for a completed research, each grounded in the report and checked for an available domain. Returns a table - name, domain, availability, price, and the other available domains for each name (the same options the website shows).
Parameters:
• researchId (string, required): UUID of the completed research to name
• target (string, optional): What to name: 'idea' (your own idea, the default) or a top bet ('bet-1'/'bet-2'/'bet-3').
list_brand_names
Read-onlyFree
List all brand-name runs generated for a specific research.
Parameters:
• researchId (string, required): Research UUID
get_brand_names
Read-onlyFree
Get one brand-name run by ID: all of its names with each name's domain, availability, price, and its other available domains.
Parameters:
• id (string, required): Brand-name run UUID
suggest_more_brand_names
Free
Append a fresh batch of names to an existing run (nothing is removed). Capped at a few rounds per run; every run render states how many rounds remain.
Parameters:
• id (string, required): Brand-name run UUID
check_brand_names_domain_availability
Free
Fill in domain availability and prices for names in a run that were saved without them (registrar throttled at generation time).
Parameters:
• id (string, required): Brand-name run UUID
delete_brand_names
DestructiveFree
Delete a brand-name run by ID. This cannot be undone.
Parameters:
• id (string, required): Brand-name run UUID
Account
check_credits
Read-onlyFree
Check your current credit balance.
- v3.3.3 Clearer description of what a report includes: grounded in live web research, with clickable sources where the data supports them.
- v3.3.0 A "Validate a startup idea" command you can pick straight from your AI app, plus setup instructions that let the assistant offer a research when you mention an idea.
- v3.2.0 Every tool now declares a title and behaviour hints, so your AI app can tell which tools only read and which ones delete.
- v3.1.0 Brand-names tools: generate name ideas, list/fetch/delete them, and check domain availability.
- v3.0.0 OAuth 2.1 replaces API keys. Connect by URL and sign in with Google - no keys to copy.
- v2.0.0 Deep-dive tools, plus retry for stuck researches and deep dives.
- v1.1.0 create_research accepts multiline ideas up to 4,000 characters.
- v1.0.0 Initial release with research tools.