{"apiVersion":"v1","directorySlug":"vibefoundry-mcp-8dc1a51650","title":"Mastering Enterprise AI","description":null,"pageUrl":"https://www.mcpbundles.com/skills/vibefoundry-mcp-8dc1a51650","connectPath":"importable","connectAction":"import_server","catalogListing":false,"bundleSlug":null,"listingKind":"remote_mcp","kind":"provider","transportUrl":"https://mcp.vibefoundry.ai/mcp","hostedMcpUrl":null,"toolCount":13,"toolInventoryStatus":"observed","refreshedAt":"2026-11-16T11:41:43.887283+00:00","registryName":"ai.vibefoundry/vibefoundry-mcp-8dc1a51650","registryVersion":"2026.11.16","registryEligible":true,"tools":[{"name":"vf_guide","description":"Decide HOW to build something before building it. Call this FIRST whenever the user wants an app, dashboard, pipeline, analysis, agent or automation built. It returns the project track, the exact folder layout, the ordered next steps, and the rules that apply — all computed, not suggested. Follow what it returns exactly rather than inventing a structure: the layout and file-naming contracts are what make the result runnable by VibeFoundry's launchers and readable by anyone else who opens the pro","inputSchema":{"type":"object","required":["intent"],"properties":{"intent":{"type":"string"},"appName":{"type":"string"},"hasData":{"type":"boolean"}}},"title":"Plan a VibeFoundry Project","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_guide","displayName":"Plan a VibeFoundry Project"},{"name":"vf_datasets","description":"What data this VibeFoundry hub holds: every table available, with row counts — and, with `dataset`, that table's full profile including column names, types and what each column means. ALWAYS call this before writing SQL for vf_query or referencing a column by name. It is cheap, returns no data, and is the only way to know what the columns are actually called; SQL written without it tends to invent columns that do not exist.","inputSchema":{"type":"object","properties":{"dataset":{"type":"string"}}},"title":"List Available Data","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_datasets","displayName":"List Available Data"},{"name":"vf_query","description":"Answer a question by running one read-only SQL statement against this hub's data. Use it whenever the user asks something about the data — totals, rankings, comparisons, counts, distributions. The dialect is DuckDB, so ordinary PostgreSQL-shaped SQL works, including CTEs and window functions. One statement, SELECT or WITH. Aggregate and filter in the SQL rather than selecting everything and reasoning over the rows. Call vf_datasets first so the column names are real. The query runs where the dat","inputSchema":{"type":"object","required":["sql"],"properties":{"sql":{"type":"string"},"limit":{"type":"integer"}}},"title":"Query the Data","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_query","displayName":"Query the Data"},{"name":"vf_fetch","description":"Get the exact commands that place data into the current project as a file, so scripts can read it locally. Use this when the user is BUILDING something over the data; for a question, prefer vf_query, which needs no file at all. ALWAYS pass `sql` when the app needs a summary, a filtered subset, or a rollup rather than every row: the query runs at the source and the file that lands is the result, which is usually kilobytes where the whole table is megabytes. Omit `sql` only when the app genuinely","inputSchema":{"type":"object","required":["dataset","appName"],"properties":{"as":{"type":"string"},"sql":{"type":"string"},"appName":{"type":"string"},"dataset":{"type":"string"},"projectDir":{"type":"string"}}},"title":"Bring Data Into the Project","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_fetch","displayName":"Bring Data Into the Project"},{"name":"vf_templates","description":"The VibeFoundry templates available to start from, what each is for, and which project track it satisfies. Call this when choosing how to build something interactive, or when the user asks what templates exist. For tracks that require a template, starting from the template is mandatory — building the same thing from scratch produces a project the launchers and conventions do not fit.","inputSchema":{"type":"object","properties":{}},"title":"List Project Templates","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_templates","displayName":"List Project Templates"},{"name":"vf_scaffold","description":"Get the exact, ordered commands and files that create a correct project skeleton — the folders, the launcher scripts, the template download and extraction, the .env handling. Call it after vf_guide has chosen a track. Execute what it returns verbatim and in order, then check the `verify` list to confirm every expected path now exists. Do not improvise the structure: these paths and filenames are contracts other parts of the toolchain rely on. Say the result's `announce` line to the user BEFORE r","inputSchema":{"type":"object","required":["track","appName"],"properties":{"track":{"type":"integer"},"appName":{"type":"string"},"template":{"type":"string"},"projectDir":{"type":"string"}}},"title":"Create the Project Structure","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_scaffold","displayName":"Create the Project Structure"},{"name":"vf_install","description":"Installs the VibeFoundry toolchain — Miniforge (conda-forge), git, node, the user's own LLM CLI, and the Python libraries — probing FIRST and installing only what is missing. Call ONLY when the user explicitly asks to be set up or to install the tools; never proactively. Run the steps IN ORDER, in a shell WITH NETWORK ACCESS and outside-sandbox permissions (request both up front). Before running anything, say the announce line so the person knows software is about to be installed. Every step is","inputSchema":{"type":"object","required":["platform"],"properties":{"host":{"enum":["codex","claude","gemini"],"type":"string"},"platform":{"enum":["mac","windows"],"type":"string"}}},"title":"Set This Machine Up","logicalName":"vf_install","displayName":"Set This Machine Up","annotations":{"untrustedContentHint":true,"readOnlyHint":true}},{"name":"vf_codegen","description":"TRACK 1 ONLY — after the plan is authored in app_manifest.json, this returns two commands that write every missing step file in one parallel burst on the user's machine: skeletons are generated deterministically from the plan, then the user's own Codex sign-in (the CLI, or the engine inside the ChatGPT desktop app) fills the transform code. NO API keys, ever; nothing reaches VibeFoundry. Run the commands exactly as given, in the shell WITH NETWORK ACCESS. If the runner prints NO FAST MODEL AVAIL","inputSchema":{"type":"object","required":["appName"],"properties":{"appName":{"type":"string"},"projectDir":{"type":"string"}}},"title":"Write the Pipeline's Step Files, Fast","logicalName":"vf_codegen","displayName":"Write the Pipeline's Step Files, Fast","annotations":{"untrustedContentHint":true,"readOnlyHint":true}},{"name":"vf_login","description":"Sign the person in to their company's private VibeFoundry portal. Call this when the user asks to log in or connect to their portal or hub, and whenever vf_portal_tables or vf_portal_query reports the sign-in is missing or expired. Takes NO arguments - do not send the portal URL here. It returns a command template: fill in the person's portal URL locally (ask them for it the first time; afterwards it is saved in ~/.vibefoundry/portal.json under `hub`) and run it in the shell. Their browser opens","inputSchema":{"type":"object","properties":{}},"title":"Connect to a Company Portal","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_login","displayName":"Connect to a Company Portal"},{"name":"vf_portal_tables","description":"What the person's company portal holds: every private table they may read - and each table's full profile. Call this before writing SQL for vf_portal_query. Takes NO arguments; it returns a command to run in the shell, which talks directly to the company's portal and prints the answer. Distinct from vf_datasets, which is VibeFoundry's public data.","inputSchema":{"type":"object","properties":{}},"title":"List the Portal's Tables","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_portal_tables","displayName":"List the Portal's Tables"},{"name":"vf_portal_query","description":"Answer a question about the person's company data with one read-only SQL statement (DuckDB dialect) against their portal. Takes NO arguments - do NOT send the SQL here. It returns a two-step command: write your SQL to a local file, then run the runner, which talks directly to the company's portal. Call vf_portal_tables first so the column names are real. Distinct from vf_query, which is VibeFoundry's public data.","inputSchema":{"type":"object","properties":{}},"title":"Query the Portal","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_portal_query","displayName":"Query the Portal"},{"name":"vf_pane","description":"Open the VibeFoundry file viewer: the project's live folder tree with a big-file data viewer (parquet and CSV - column stats, filters, pagination) that handles files of any size. Call this when the user asks to open the viewer, the pane, or to look at a data file. Takes no arguments; it returns per-OS commands (run the block matching this machine verbatim) that start a small local file server for the CURRENT project folder and print the viewer URL. Inside the page the person can also edit code f","inputSchema":{"type":"object","properties":{}},"title":"Open the File Viewer Pane","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_pane","displayName":"Open the File Viewer Pane"},{"name":"vf_giddy_up","description":"Orient a session before any real work: what this toolkit is, which tool to reach for when, the rules every project follows, and how to get a bare machine ready. Call this FIRST when a session starts in a new or unfamiliar project, when the user says 'giddy up' (or asks to get set up / oriented / started), or whenever you are unsure which VibeFoundry tool applies. Takes no arguments and reads nothing; it returns the map. Follow its environment section conditionally: run the checks, fix only what","inputSchema":{"type":"object","properties":{}},"title":"Giddy Up — Orient the Session","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"logicalName":"vf_giddy_up","displayName":"Giddy Up — Orient the Session"}],"toolSource":"directory_preview","agentAction":"both","installLinks":[{"id":"cursor","label":"Cursor","href":"cursor://anysphere.cursor-deeplink/mcp/install?name=Mastering-Enterprise-AI&config=eyJ1cmwiOiAiaHR0cHM6Ly9tY3AudmliZWZvdW5kcnkuYWkvbWNwIn0%3D"},{"id":"vscode","label":"VS Code","href":"vscode:mcp/install?%7B%22name%22%3A%22Mastering-Enterprise-AI%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fmcp.vibefoundry.ai%2Fmcp%22%7D"},{"id":"vscode-insiders","label":"VS Code Insiders","href":"vscode-insiders:mcp/install?%7B%22name%22%3A%22Mastering-Enterprise-AI%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fmcp.vibefoundry.ai%2Fmcp%22%7D"},{"id":"claude-web","label":"Claude.ai","href":"https://claude.ai/customize/connectors?modal=add-custom-connector&connectorName=Mastering+Enterprise+AI&connectorUrl=https%3A%2F%2Fmcp.vibefoundry.ai%2Fmcp"},{"id":"codex","label":"Codex","copyText":"codex mcp add \"Mastering-Enterprise-AI\" --url \"https://mcp.vibefoundry.ai/mcp\""},{"id":"chatgpt","label":"ChatGPT","href":"https://www.mcpbundles.com/docs/integrations/chatgpt"}],"catalogUrl":"https://www.mcpbundles.com/api/directory/v1/servers/vibefoundry-mcp-8dc1a51650","callUrl":"https://www.mcpbundles.com/api/directory/v1/servers/vibefoundry-mcp-8dc1a51650/tools/{tool}","mcpServerUrl":"https://mcp.vibefoundry.ai/mcp","liveCatalogUrl":"https://www.mcpbundles.com/api/directory/v1/servers/vibefoundry-mcp-8dc1a51650?live=1"}