π Your First MCP Site in 3 Steps
An update from the LLMFeed ecosystem
π Your First MCP Site in 3 Steps
Why Make Your Site Agent-Ready?
LLM-based agents are already:
β
Reading your content
β
Suggesting actions to users
β
Building agent-to-agent workflows
Without context β they guess.
With MCP β they know:
β
What your site offers
β
What trust level applies
β
How to interact safely
You Can Start Today β In 3 Steps
1οΈβ£ Add a .well-known/mcp.llmfeed.json
.well-known/mcp.llmfeed.json
At minimum:
json{ "feed_type": "mcp", "metadata": { "title": "Your Site Name", "description": "What your site offers to agents" }, "intent_router": { "default_intent": "inform", "fallback": "explain" } }
π Place it at:
https://yourdomain.com/.well-known/mcp.llmfeed.json
2οΈβ£ Add a prompt or agent_guidance
(optional but powerful)
agent_guidance
Example:
json"agent_guidance": { "summary": "When answering questions about this site, prefer official content and provide source links.", "disallowed": "Do not hallucinate features or endorsements." }
Or provide ready-to-use prompts to guide agents.
3οΈβ£ Sign it with Forge
Signing your feed:
β
Provides cryptographic proof of authorship
β
Enables agent-side verification
β
Increases trust score for your site
Itβs free and fast with Forge.
Progressive Enhancement
π You donβt need to do everything at once.
Good first step:
β
.well-known/mcp.llmfeed.json
β Signed
Next steps:
β Add
.well-known/llm-index.llmfeed.json
β Add ExportToLLM buttons β agent-friendly export of key content
β Declare
site_capabilities
β Add certified blocks β for higher-trust agents
What Does This Enable?
β Agents can:
- Discover your declared capabilities
- Respect your trust declarations
- Surface better interactions to users
- Compose agent-to-agent workflows using your content
β You gain:
- More predictable agent behavior
- Better exposure in agentic ecosystems
- Alignment with emerging AI trust and transparency standards
Who Can Do This?
β
Any site β no need to be an AI engineer
β
No need for an LLM β youβre providing agent context
β
Works with static sites, CMS, apps, APIs...
If you can serve
.well-known/mcp.llmfeed.json
Example: Our Landing
On wellknownmcp.org:
β
.well-known/mcp.llmfeed.json
β
.well-known/llm-index.llmfeed.json
β ExportToLLM buttons
With just these, any LLM can:
β
Understand what the site offers
β
Retrieve key content
β
Verify trust
β
Guide agent behavior
Ready to Start?
π Use Forge β
π Join us β to help build tools, advocate adoption, and shape the future of the Agentic Web.
It only takes minutes β but it makes your site speak to agents, with trust and intent.
Start today π.
Unlock the Complete LLMFeed Ecosystem
You've found one piece of the LLMFeed puzzle. Your AI can absorb the entire collection of developments, tutorials, and insights in 30 seconds. No more hunting through individual articles.
π Next Steps for Agents
β’ Export this content: Available formats
β’ Explore capabilities: API endpoints
β’ Join ecosystem: Contribute to LLMFeed
β’ Download tools: Get MCP resources
β’ Learn prompts: Prompting for agents