API and agent access

Name: Asset Forge Index dataset
Creator: Asset Forge Index

Everything on Asset Forge Index is available as static, read-only files: no key, no sign-up. Use it from an LLM agent, a script or a spreadsheet. Please link back to the tool page when you cite data.

Endpoints

/llms.txt	Short site guide for LLMs (llmstxt.org format)
/llms-full.txt	All tools in one Markdown file
/api/tools.json	All tools, full records
/api/tools/<slug>.json	One tool, for example /api/tools/meshy.json
/tools/<slug>.md	One tool as Markdown, for example /tools/meshy.md
/api/changes.json	Dated changelog and tracked page changes, newest first
/feed.xml	Atom feed of changes
/data/tools.csv	Flat CSV
/api/schema.json	JSON Schema of a tool record
/api/openapi.json	OpenAPI 3.1 description (usable as a GPT action or tool spec)
/api/index.json	Endpoint and slug index

Field rules

null means the vendor page doesn't state it. "unverified" means we couldn't confirm it.
paid_price.display is copied exactly as shown on the vendor page, including billing period and promos.
commercial_use.free / .paid is our one-word summary (yes, yes_with_attribution, conditional, no, unverified, not_stated, n/a). Always read license and source_urls before relying on it.
schema_version is currently 1.0. Fields may be added without notice. Breaking changes would ship under /api/v2/.
Data is regenerated weekly; last_checked is per tool.

Example

curl -s https://assetforgeindex.com/api/tools.json | jq '.tools[] | select(.commercial_use.free=="yes") | .name'

{
  "slug": "meshy",
  "name": "Meshy",
  "free_tier": "Free: 100 monthly credits, 1 concurrent task",
  "license": {
    "free": "CC BY 4.0 (commercial use with attribution)",
    "paid": "Private licence; assets 'Private & Customer owned'"
  },
  "commercial_use": {
    "free": "yes_with_attribution",
    "paid": "yes",
    "basis": "Editorial summary of the licence text in license.free / license.paid"
  },
  "last_checked": "2026-10-15"
}

Hosted MCP server

Endpoint: https://assetforgeindex.com/mcp (MCP Streamable HTTP, JSON-RPC over POST). Read-only, stateless, no key and no sign-up. It speaks MCP 2026-07-28 (per-request metadata, server/discover) and the 2025 revisions (initialize handshake, 2025-03-26 to 2025-11-25).

`search_tools`	`query`, optional `category` (2d, 3d, audio, coding, prompt-to-game), optional `commercial_free` (true/false), `limit`
`get_tool`	`slug`: the full record for one tool
`compare`	`slugs`: 2 to 6 tools side by side
`recent_changes`	optional `since` (YYYY-MM-DD), `limit`

Every result includes citations (vendor source URL and the date we checked it) and the dataset date, both in structuredContent and in the text block, so agents can cite the vendor page.

Claude (Desktop, web, mobile)

Customize › Connectors › Add custom connector, then paste https://assetforgeindex.com/mcp. No authentication is needed. Or, for a local stdio bridge, add this to claude_desktop_config.json:

{
  "mcpServers": {
    "asset-forge-index": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://assetforgeindex.com/mcp"
      ]
    }
  }
}

Cursor

Add to ~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "asset-forge-index": {
      "url": "https://assetforgeindex.com/mcp"
    }
  }
}

Codex

Run codex mcp add asset-forge-index --url https://assetforgeindex.com/mcp, or add to ~/.codex/config.toml:

[mcp_servers.asset-forge-index]
url = "https://assetforgeindex.com/mcp"

Test it with curl

curl -s https://assetforgeindex.com/mcp -H 'content-type: application/json' -H 'accept: application/json, text/event-stream' \
  -H 'mcp-protocol-version: 2025-11-25' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"search_tools","arguments":{"query":"pixel","commercial_free":true}}}'

Using it without MCP

Any agent that can fetch URLs can use this data. Point it at https://assetforgeindex.com/llms.txt, or give it https://assetforgeindex.com/api/openapi.json as a tool definition.

Crawlers

AI and search crawlers are welcome; see robots.txt.