General Tools

aivoicecall.net

An MCP server for browsing a portfolio of demos, checking their health, learning about the author, and creating live diagram sessions.

5 endpoints24 known toolsFirst detected September 14, 2026Last detected September 14, 2026

ENDPOINT 1

https://aivoicecall.net/mcp

No auth detected

MCP server metadata

Name: portfolio
Version: 0.1.0

Capabilities

experimentalpromptsresourcestools

Server instructions

A directory of Thai Nguyen's live demo projects — eleven working applications, each on its own public URL. Start with list_demos() to see what exists; it returns each demo's pitch, what it deliberately does NOT do, and — importantly — its own MCP endpoint where it has one, so you can connect to a demo directly and use it rather than read about it. get_demo(id) gives the full detail on one. fleet_health() checks every public URL right now and returns response times, so you can verify the claims instead of trusting them. about_the_author() has the background and how to make contact. Everything here is read-only and there is nothing to authenticate.

Known tools 4

list_demos

Every demo in the portfolio: id, name, one-line pitch, public URL, and its own MCP endpoint where it has one.

Inferred read-only

get_demo

Everything about one demo: what it does, what it is for, what it deliberately does NOT do, its features and headline numbers, and how to drive it from an agent.

Inferred read-only

fleet_health

Check every demo's public URL right now and report which are reachable, with response times.

Inferred read-only

about_the_author

Who built this: background, current focus, location, and how to get in touch.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.portfolio]
url = "https://aivoicecall.net/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "portfolio": {
      "type": "http",
      "url": "https://aivoicecall.net/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: portfolio
Remote MCP URL: https://aivoicecall.net/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "portfolio": {
      "url": "https://aivoicecall.net/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "portfolio": {
      "type": "http",
      "url": "https://aivoicecall.net/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "portfolio",
  "transport": "streamable-http",
  "url": "https://aivoicecall.net/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

ENDPOINT 2

https://diagram.aivoicecall.net/mcp

No auth detected

MCP server metadata

Name: flowmind
Version: 1.0.0

Capabilities

experimentalpromptsresourcestools

Server instructions

Create and edit live Mermaid diagrams that a human watches render in a browser tab. Typical flow: call create_diagram(prompt=...) or create_diagram(code=...) to start one — it returns an id and a URL; give the URL to the person so they can open it. Then update_diagram(id, code) pushes a new version and their tab re-renders instantly. get_diagram(id) returns what is on screen RIGHT NOW — the current source, the version, and whether it rendered (with the exact Mermaid error if the person's own edit broke it), so you can read their changes and fix syntax errors. generate_diagram(prompt, id) turns a plain-language request into Mermaid with the model (into an existing session, or a new one). The id is the capability: whoever holds it can read and edit the diagram. Always output the COMPLETE diagram when you update — never a fragment.

Known tools 4

create_diagram

Start a new live diagram session and return its id and shareable URL.

Inferred read-only

generate_diagram

Turn a plain-language request into a Mermaid diagram with the model.

Inferred read-only

get_diagram

Return what is on screen for this session RIGHT NOW: the current Mermaid source, the version number, and whether it renders — including the exact Mermaid error text if the person's own browser edit broke the syntax.

Potential side effects

update_diagram

Replace the diagram's source with `code` (the COMPLETE Mermaid, never a fragment) and push it to the person's open browser tab, which re-renders instantly.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.flowmind]
url = "https://diagram.aivoicecall.net/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "flowmind": {
      "type": "http",
      "url": "https://diagram.aivoicecall.net/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: flowmind
Remote MCP URL: https://diagram.aivoicecall.net/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "flowmind": {
      "url": "https://diagram.aivoicecall.net/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "flowmind": {
      "type": "http",
      "url": "https://diagram.aivoicecall.net/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "flowmind",
  "transport": "streamable-http",
  "url": "https://diagram.aivoicecall.net/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

ENDPOINT 3

https://webscrape.aivoicecall.net/mcp

No auth detected

MCP server metadata

Name: webscrape
Version: 0.1.0

Capabilities

experimentalpromptsresourcestools

Server instructions

A live demo that turns a web page into structured rows. Point it at a page with a list on it — a catalogue, a directory, search results — and say what you want in plain English; it reads the page (operating search forms when it has to), extracts the rows, and hands them back as JSON. It ships pre-built recipes for public scraping sandboxes (books/quotes.toscrape.com, scrapethissite.com, Hacker News). Start with list_recipes to see the ready-made extractors, or analyze_page(url) to inspect any page first. run_recipe(id) is the fast, deterministic path; extract(url, prompt) is the general one. Both run to completion and return the rows; a long run hands back a run_id you can poll with get_run. Contact fields are masked at storage time and results are capped at 100 rows. Real scrape runs are rate-limited per visitor because this is an open demo with no sign-in; a refusal says so plainly and is not an error. Only public URLs work — private/internal hosts are refused by an SSRF guard.

Known tools 6

list_recipes

List the pre-built, reusable extractors (recipes) this demo ships — each with the page it targets and what it pulls.

Inferred read-only

list_runs

List recent extraction runs, newest first (run_id, url, status, row_count).

Inferred read-only

get_run

Get one run's status and its extracted rows (already masked).

Inferred read-only

analyze_page

Inspect a URL WITHOUT extracting: report the detected page type and the smartest extraction strategy (e.g.

Inferred read-only

run_recipe

Run a saved recipe to completion and return its extracted rows.

Inferred read-only

extract

The headline verb: turn any public page into structured JSON rows.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.webscrape]
url = "https://webscrape.aivoicecall.net/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "webscrape": {
      "type": "http",
      "url": "https://webscrape.aivoicecall.net/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: webscrape
Remote MCP URL: https://webscrape.aivoicecall.net/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "webscrape": {
      "url": "https://webscrape.aivoicecall.net/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "webscrape": {
      "type": "http",
      "url": "https://webscrape.aivoicecall.net/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "webscrape",
  "transport": "streamable-http",
  "url": "https://webscrape.aivoicecall.net/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

ENDPOINT 4

https://aigateway.aivoicecall.net/mcp

No auth detected

MCP server metadata

Name: aigateway
Version: 0.1.0

Capabilities

experimentalpromptsresourcestools

Server instructions

A live shared LLM/AI gateway that routes chat, vision, embeddings, speech, image and music across cloud APIs, self-hosted GPU servers and rented cards, with failover and cost tracking. This interface is READ-ONLY observability. Use list_models() to see what the gateway can route (the modality x lane capability matrix), cost_report() for spend and the savings versus a single paid provider, failover_state() for which providers are up right now in routing-priority order, and health_history() for uptime and latency over a window. It deliberately does NOT expose chat, embeddings, image or audio generation — the caller is already a model, and the gateway's job is to route models, not to be one more. Every number is the aggregate, redacted public view: no endpoint URLs, no keys, no machine names. Rate-limited per visitor.

Known tools 4

list_models

What the gateway can route right now: the modality x lane capability matrix (which modalities are served on cloud, self-hosted, container, native, rented lanes) plus the live model list with capability, context window and daily quota.

Inferred read-only

cost_report

Aggregate traffic (request counts, average latency, success rate) and the headline savings: what the same tokens would have cost on one paid provider versus what they actually cost across the gateway's lanes.

Inferred read-only

failover_state

Every provider's current health and routing priority, highest-priority first — the live failover posture.

Inferred read-only

health_history

Uptime (fraction of probes healthy) and average latency per provider over a recent window, with a downsampled tail of probe points.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.aigateway]
url = "https://aigateway.aivoicecall.net/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "aigateway": {
      "type": "http",
      "url": "https://aigateway.aivoicecall.net/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: aigateway
Remote MCP URL: https://aigateway.aivoicecall.net/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "aigateway": {
      "url": "https://aigateway.aivoicecall.net/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "aigateway": {
      "type": "http",
      "url": "https://aigateway.aivoicecall.net/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "aigateway",
  "transport": "streamable-http",
  "url": "https://aigateway.aivoicecall.net/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

ENDPOINT 5

https://aiagent.aivoicecall.net/mcp

No auth detected

MCP server metadata

Name: aiagent
Version: 0.1.0

Capabilities

experimentalpromptsresourcestools

Server instructions

A live demo knowledge base. Drop in a public web page, wait for it to be indexed, then ask questions and get answers with citations back to the source passages. The library is shared and starts nearly empty, so if a search returns nothing, call list_documents before concluding the answer is not there — an empty library and a missing answer look identical otherwise. ingest_url is asynchronous. It returns while the page is still being parsed, summarised and embedded; poll ingestion_status until it reports 'ready' before searching for what you just added. Only public web addresses can be ingested, and answers are limited per visitor per hour because this is an open demo with no sign-in.

Known tools 6

search_knowledge

Search the library and return the best-matching passages with relevance scores.

Inferred read-only

list_documents

List every document in the library with its ingest status.

Inferred read-only

read_document

Read the extracted text of one document, by the id returned from list_documents, search_knowledge or a citation.

Inferred read-only

ingestion_status

Check whether a document has finished indexing.

Inferred read-only

ask

Ask a question and get an answer grounded in the library, with citations back to the passages it used.

Inferred read-only

ingest_url

Add a public web page to the library.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.aiagent]
url = "https://aiagent.aivoicecall.net/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "aiagent": {
      "type": "http",
      "url": "https://aiagent.aivoicecall.net/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: aiagent
Remote MCP URL: https://aiagent.aivoicecall.net/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "aiagent": {
      "url": "https://aiagent.aivoicecall.net/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "aiagent": {
      "type": "http",
      "url": "https://aiagent.aivoicecall.net/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "aiagent",
  "transport": "streamable-http",
  "url": "https://aiagent.aivoicecall.net/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

TRUST AND VERIFICATION EVIDENCE

Loading Trust v2 evidence…

Checking the associated registrable domain. The BuiltWith key remains server-side.

Indexed

Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.