Reference Links
Reference Links
Canonical standards, docs, specs, and source links for agent-readable software surfaces.
This is the single source index for standards, specs, and official docs that Agent Surface tracks. Use it when you want the primary source rather than the curated tooling view.
| Reference | What It Covers | Use It For |
|---|
| Cloudflare Agent Readiness | Discoverability, content accessibility, bot access control, and capabilities | Understanding what agent-readiness scanners currently check. |
| Agent Readiness Checker | Public site scanner for agent readiness | Spot-checking whether a URL exposes agent-readable signals. |
| Markdown for Agents | Accept: text/markdown, x-markdown-tokens, Content-Signal, and edge HTML-to-Markdown conversion | Serving readable pages to agents without wasting context on HTML. |
| llms.txt | LLM-readable site indexes | Publishing a concise map of docs, APIs, and key resources. |
| Content Signals | Search, AI input, and AI training policy signals | Declaring how accessed content may be used. |
| Web Bot Auth | Signed HTTP requests for bot identity | Letting sites distinguish accountable agents from anonymous automation. |
| API Catalog RFC 9727 | /.well-known/api-catalog | Advertising API specs, docs, status pages, and related API metadata. |
| Reference | What It Covers | Use It For |
|---|
| Agent Skills | Open format for reusable folders of instructions, scripts, references, and assets | Packaging repeatable agent capabilities without bloating the base prompt. |
| Agent Skills Specification | SKILL.md frontmatter, directories, validation, and progressive disclosure | Authoring skills that portable agents can discover and load. |
| Adding Skills Support | Skill discovery, catalogs, activation, and context management | Implementing a skills-compatible agent runtime or client. |
| Agent Skills Reference Library | Validation and prompt XML generation tooling | Validating local or published skills. |
| AGENTS.md | Repository instructions for coding agents | Giving agents repo-level commands, boundaries, conventions, and navigation. |
| Reference | What It Covers | Use It For |
|---|
| MCP Introduction | MCP as the standard connector between AI apps and external systems | Grounding protocol decisions. |
| MCP Architecture | Clients, servers, hosts, and capability flow | Designing where MCP belongs in an agent system. |
| Build an MCP Server | Server implementation basics | Exposing tools, resources, and prompts to agents. |
| Connect to Remote MCP Servers | Remote server connection model | Publishing MCP over network transport. |
| MCP Inspector | Interactive server inspection and debugging | Testing tool definitions, schemas, prompts, and resources. |
| MCP Registry | Public MCP server discovery | Finding and distributing MCP servers. |
| Reference | What It Covers | Use It For |
|---|
| OAuth Protected Resource Metadata RFC 9728 | /.well-known/oauth-protected-resource | Letting agents discover authorization servers and resource metadata. |
| OAuth 2.1 Draft | Modern OAuth profile | Designing delegated machine-to-machine auth. |
| DPoP RFC 9449 | Proof-of-possession access tokens | Reducing bearer-token replay risk for sensitive tools. |
| Reference | What It Covers | Use It For |
|---|
| Mastra Docs | TypeScript agents, tools, workflows, memory, RAG, MCP, evals, and voice | Default local TypeScript agent scaffolding. |
| Vercel AI SDK | Model calls, tools, structured output, streaming, and UI | TypeScript AI app foundations. |
| OpenAI Agents SDK | Agents, tools, handoffs, guardrails, tracing, and evals | OpenAI-native agent systems. |
| OpenRouter Docs | Model routing, structured outputs, plugins, and hosted tools | Provider optionality and first-party model-call tools. |
| Cloudflare Agents | Durable Object-backed agents, state, scheduling, tools, MCP, and browser access | Workers-native agent systems. |
| Cloudflare AI Gateway | Gateway observability, caching, rate limits, guardrails, retries, and routing | Governed model access across providers. |
| Cloudflare Browser Run | Managed browser automation, screenshots, PDFs, Markdown extraction, crawling, Playwright, and MCP | Browser access for agents. |
| Cloudflare Sandbox | Isolated code execution for agents and applications | Running untrusted or generated code outside the app process. |
| Reference | What It Covers | Use It For |
|---|
| x402 | HTTP-native machine payments | Pay-to-access resources and APIs. |
| Universal Commerce Protocol | Commerce discovery and transaction protocol | Agent-driven ecommerce surfaces. |
| Agentic Commerce Protocol | Agent-mediated purchasing flows | Agent-initiated buying with explicit payment boundaries. |
Use this page for canonical sources and specs. Use the Tooling Catalog for curated recommendations, tiers, and judgement calls.