{
 "url": "https://agenticweb.wiki/agentic-web/",
 "title": "What is the agentic web? · agenticweb.wiki",
 "description": "The agentic web is the ordinary web plus conventions that let software agents discover, identify, read, act, coordinate, pay and be held to account.",
 "markdown_url": "https://agenticweb.wiki/agentic-web/index.md",
 "markdown": "The agentic web is the ordinary HTTP/DNS web extended with machine-native discovery files, signed identity headers, permission tokens, and invocation or payment protocols that let software agents discover, authenticate, act, and transact without scraping human UI. As of 2026-10-30, agenticweb.wiki maps 132 standards across 9 lifecycle steps (Discover, Identify, Permit, Read, Act, Coordinate, Pay, Present, Account) and 325 verified implementations.\n\n# Agentic web\n\nThe agentic web is the ordinary web plus a layer of conventions that let software agents — rather than people driving browsers — discover what a site offers, prove who they are, learn what they may do, read its content, invoke its capabilities, coordinate with other agents, pay for what they use, present results to a person, and be held to account afterwards. None of this replaces HTTP, HTML or DNS. It sits on top of them: a set of well-known files, endpoints, headers and cards that a site can add so an agent does not have to guess. This site maps that layer as nine steps and treats every claim about it as something to source or mark as unverified, not something to assume.\n\n## What \"agentic\" adds\n\nThe conventional web was built for a person with a browser: pages carry visual layout, forms expect a human to fill them in, and a click is the unit of action. An agent arriving at the same URL has none of a browser's affordances to fall back on unless it renders the page like one, which is slow, brittle, and easy to break with a redesign. What \"agentic\" adds is a parallel, machine-native surface: a /.well-known/agent-card.json that states an A2A endpoint and its skills instead of a \"Contact us\" button; an MCP endpoint that exposes tools with typed schemas instead of a search box; a robots.txt with separate crawler tokens for search inclusion, model training and user-triggered fetches instead of one blanket disallow. The word does not describe a new network or a new browser. It describes a site choosing to publish, in a form an agent can parse without inference, the same things it already publishes for people.\n\n## What it is not\n\nAgentic AI is a capability: a model that plans across multiple steps and calls tools to get something done. It can exist entirely inside one vendor's walls — a coding assistant editing files, a customer-support bot querying an internal database — with no reference to any other site's protocols at all. The agentic web is the opposite half of the picture: it is what a site publishes so that somebody else's agent, which it does not control and may never have heard of, can find and use it correctly. An agent can be highly capable and interact with zero agentic-web infrastructure; a site can publish a flawless Agent Card and still be talked to by a very simple bot.\n\nThe Semantic Web proposed a universal, machine-readable ontology of meaning — RDF triples, OWL classes, a web where any two systems could in principle reason about the same facts. The agentic web makes no such bet. Its protocols are imperative and narrow: fetch this card, call this endpoint, sign this request, settle this payment. There is no shared ontology requirement and no expectation that an agent \"understands\" a site beyond what a card or schema tells it to do next.\n\nWeb3 is a project about decentralised ownership and settlement, chiefly through blockchains. The agentic web borrows blockchain rails where they are useful for machine-to-machine payment — x402 and several of its adjacent proposals settle in stablecoins, and identity proposals such as ERC-8004 anchor agent registries on-chain — but most of the agentic web's standards layer (discovery files, agent cards, robots directives, HTTP signatures) has nothing to do with decentralisation and runs on ordinary client-server infrastructure that any single company can operate alone.\n\nBrowser automation is an agent pretending to be a person: it opens a real browser, reads the rendered DOM or accessibility tree, and clicks buttons and fills forms designed for human hands. It works against any site, including ones that have done nothing to accommodate agents, but it is slow, breaks when a layout changes, and gives the site no way to distinguish a considerate agent from an adversarial scraper. The agentic web is the alternative a site can offer instead: a native interface — an MCP tool, an A2A skill, a signed request — that does not require pretending to be a browser at all. WebMCP sits at the boundary between the two: it lets a page register tools that an agent embedded in the visitor's own browser can call directly, rather than having that agent click through the rendered UI, but as of Chrome's origin-trial rollout announced 9 June 2026 it is still experimental and origin-scoped, not a universally available browser capability.\n\n## The nine steps\n\n| # | Step | Question | Classic-web analog | Standards |\n| --- | --- | --- | --- | --- |\n| 1 | [Discover](/category/discover/) | What is here, and where? | sitemap, .well-known | 18 |\n| 2 | [Identify](/category/identify/) | Who is this agent? Who is this site? | TLS certificate, WHOIS | 22 |\n| 3 | [Permit](/category/permit/) | What may the agent do here? | robots.txt, terms of use | 32 |\n| 4 | [Read](/category/read/) | Give me the content | HTML, RSS | 6 |\n| 5 | [Act](/category/act/) | Do something for me | forms, REST APIs | 13 |\n| 6 | [Coordinate](/category/coordinate/) | Talk to another agent | none; agent-to-agent is new | 13 |\n| 7 | [Pay](/category/pay/) | Settle the bill | HTTP 402, finally used | 17 |\n| 8 | [Present](/category/present/) | Show a human the result | the browser | 4 |\n| 9 | [Account](/category/account/) | Prove what happened; be judged | logs, audit trails | 7 |\n\nAn agent's path across a site can be described independently of any one protocol: it has to discover what is there, identify itself and the site, learn what it is permitted to do, read content, invoke a capability, coordinate with other agents where more than one is involved, pay if the resource is not free, present the outcome to a person, and leave a record that can be checked afterwards. This site organises every standard it tracks under one of these nine steps; each has a classic-web analogue that most of the concepts descend from.\n\n## How the pieces fit together\n\nThe steps are not independent of each other in practice. Discovery and identity typically arrive together: an A2A client fetches a site's Agent Card at /.well-known/agent-card.json (itself a use of the well-known-URI convention that RFC 8615 reserved in May 2019) and the card's own content — supported authentication schemes, provider identity — is what lets the client decide whether to trust what it just found. Permission sits alongside both: a robots.txt with distinct tokens for search crawling, training and user-triggered fetches (OpenAI's OAI-SearchBot, GPTBot and ChatGPT-User are documented as three separate policies, and Anthropic and Perplexity publish an equivalent three-way split) tells an agent what it may do with what it has found before it acts on it.\n\nInvocation then depends on what discovery handed over: an MCP client that has parsed a server's tools uses that same session to call them; an A2A client that has read a skill's input/output modes structures its request accordingly. Payment, where it appears, is a further step layered on top of invocation rather than a replacement for it — Google's Agent Payments Protocol (AP2) chains an intent mandate, a cart mandate and a payment mandate as cryptographically signed, scoped authorisations so a person can let an agent transact without handing it standing account access, while x402 instead resumes an HTTP request that received a 402 status by attaching a signed payment payload, optionally verified and settled by a third-party facilitator rather than the resource server itself. Both approaches assume identity and permission were already established upstream; neither one does discovery's job for it.\n\n## What is shipping today\n\nCoverage varies sharply by step. The Model Context Protocol, published by Anthropic and now governed under the Agentic AI Foundation and the Linux Foundation, is implemented by multiple independent clients — Claude Desktop, Cursor and Zed among them — against its current 2026-07-28 specification, and is the protocol behind the OpenAI Apps SDK's tool ecosystem. The Agent2Agent Protocol (A2A), originated by Google and now under Linux Foundation governance, reports a stable 1.0.x line with adoption named by Google's own Agent Development Kit, by Microsoft's Agent Framework (from 28 April 2026) and by Microsoft Work IQ (from 21 August 2026). x402, from Coinbase and now stewarded by an x402 Foundation, has multiple independent facilitator services running in production across several blockchain networks handling verification and settlement on a resource server's behalf, alongside servers that \"self-facilitate\" the same logic in-process. Web Bot Auth's signed-request approach to bot identification has separate documented implementations from Google, Cloudflare and Vercel. AP2 launched with more than sixty named partner organisations. Against that, most of the other protocols this site tracks — the majority of the discovery and coordinate rows in particular — have a specification and, at most, one implementing organisation; this site marks that distinction on every protocol page rather than treating \"has a GitHub repository\" as evidence of use.\n\n## Classic web to agentic web translation\n\nEvery standard on this map replaces or extends a classic web mechanism that breaks when the visitor is an autonomous software agent rather than a person in a browser. The table below is compiled directly from each record's implementation tier ([IETF RFC 7942 / W3C running-code bar](/methodology/)) and live /check/ probe counts:\n\n| Classic web primitive | Step | Why it breaks for agents | Multi-implementation & shipping in production | Trial, prototype & paper drafts |\n| --- | --- | --- | --- | --- |\n| sitemap.xml & OpenAPI | [1 Discover](/category/discover/) | Lists human URLs or raw REST routes without agent skills or capability metadata | [api-catalog (RFC 9727)](/protocols/api-catalog-rfc9727/) (multi-implementation), [MCP Server Cards](/protocols/mcp-server-cards/) (multi-implementation), [ARD](/protocols/ard/) (multi-implementation) | [agents.txt](/protocols/agents-txt/) (prototype), [agents.json](/protocols/agents-json/) (prototype), [agent-permissions.json](/protocols/agent-permissions-json/) (prototype), [A Framework for Agent Discovery in DAWN](/protocols/dawn-agent-discovery-framework/) (none found) |\n| DNS A/SRV & WHOIS | [1 Discover](/category/discover/) | Resolves hostnames to IP addresses, not agent endpoints or cryptographic keys | [AID](/protocols/aid/) (shipping), [Agent Directory Service](/protocols/ads/) (shipping · 1/1730 origins), [OASF](/protocols/oasf/) (shipping), [Agent Name Service v2 (ANS)](/protocols/ans/) (shipping) | [DNS-AID](/protocols/dns-aid/) (prototype), [Discovery of Model Context Protocol Servers via DNS TXT Records](/protocols/mcp-dns-discovery/) (none found) |\n| TLS certs & User-Agent strings | [2 Identify](/category/identify/) | User-Agent is trivially spoofed; IP allowlists break across cloud and residential proxies | [HTTP Message Signatures](/protocols/http-message-signatures/) (multi-implementation), [Web Bot Auth](/protocols/webbotauth/) (multi-implementation), [SPIFFE](/protocols/spiffe/) (multi-implementation), [ERC-8004: Trustless Agents](/protocols/erc-8004/) (multi-implementation), [The Agent Identity Document: A Hosted, Accountable Public Record for AI Agents](/protocols/agent-identity-document/) (shipping) | [Signature Agent Card (Web Bot Auth registry)](/protocols/web-bot-auth-signature-agent-card/) (prototype), [WIMSE](/protocols/wimse/) (prototype) |\n| OAuth 2.0 consent screens | [3 Permit](/category/permit/) | Assumes a human clicking 'Allow' in an interactive browser redirect loop | [OAuth Protected Resource and Authorization Server Metadata (as used for agent authorization)](/protocols/oauth-metadata-for-agents/) (multi-implementation · 114/1730 origins), [CIMD](/protocols/oauth-cimd/) (multi-implementation), [auth.md](/protocols/auth-md/) (multi-implementation), [ID-JAG](/protocols/identity-assertion-authz-grant/) (multi-implementation) | [Agent Authorization Grant](/protocols/aauth/) (none found), [DAAP](/protocols/daap/) (prototype), [Agent Authorization Profile](/protocols/aap/) (prototype), [AIMS](/protocols/wimse-aims/) (none found) |\n| robots.txt (Allow/Disallow) | [3 Permit](/category/permit/) | Binary crawl rules cannot separate search indexing, live RAG grounding, and model training | [AI crawler tokens in robots.txt](/protocols/ai-crawler-tokens-robots/) (multi-implementation), [Content Signals Policy](/protocols/cloudflare-content-signals-policy/) (multi-implementation), [Really Simple Licensing](/protocols/rsl/) (shipping · 4/1730 origins), [TDMRep](/protocols/tdmrep/) (shipping) | [AIPREF](/protocols/aipref-vocabulary/) (none found), [AIPREF](/protocols/aipref-content-usage-attachment/) (prototype) |\n| HTML DOM & RSS feeds | [4 Read](/category/read/) | Layout boilerplate, navbars, and client JS waste 80%+ of context tokens | [llms.txt](/protocols/llms-txt/) (multi-implementation), [Markdown for Agents](/protocols/cloudflare-markdown-for-agents/) (shipping), [AGENTS.md](/protocols/agents-md/) (multi-implementation), [Open Knowledge Format](/protocols/okf/) (shipping), [Agent Readability Spec](/protocols/vercel-agent-readability/) (shipping) | [Ossie](/protocols/osi/) (prototype), [The 'knowledge-linkset' Well-Known URI for Publishing Knowledge Artefacts](/protocols/knowledge-linkset/) (none found) |\n| HTML forms & bespoke REST APIs | [5 Act](/category/act/) | Every site has custom form fields; browser automation breaks when UI layouts change | [Model Context Protocol](/protocols/mcp/) (multi-implementation), [Agent Skills](/protocols/anthropic-agent-skills-format/) (multi-implementation), [Agent Client Protocol](/protocols/agent-client-protocol/) (multi-implementation), [OpenAI Apps SDK manifests](/protocols/openai-apps-sdk-manifests/) (shipping), [NLWeb](/protocols/nlweb/) (multi-implementation) | [WebMCP](/protocols/webmcp/) (prototype) |\n| Webhooks & microservice RPC | [6 Coordinate](/category/coordinate/) | Built for pre-integrated internal services, not cross-organization task delegation | [Agent2Agent Protocol (A2A)](/protocols/a2a/) (multi-implementation), [Agent Network Protocol](/protocols/anp/) (shipping), [SLIM](/protocols/slim/) (shipping), [Coral Protocol (CoralOS)](/protocols/coral-protocol/) (shipping), [AAMP](/protocols/aamp/) (multi-implementation) | [Agent-Channel Frame for Identity-Keyed Fan-Out Delivery to Concurrent Sessions](/protocols/agent-channel-fan-out/) (none found) |\n| HTTP 402 (unused since 1997) | [7 Pay](/category/pay/) | API keys and subscriptions require human signup, credit cards, and monthly billing | [x402](/protocols/x402/) (multi-implementation), [MPP](/protocols/mpp/) (multi-implementation), [L402](/protocols/l402/) (shipping), [x401: HTTP Proof Requirement Protocol](/protocols/x401/) (shipping) | [Cloudflare Pay Per Crawl](/protocols/cloudflare-pay-per-crawl/) (prototype) |\n| 3-D Secure & card checkout | [7 Pay](/category/pay/) | Flags 'human-not-present' agent purchases as bot fraud; no cryptographic proof of user budget | [Agentic Commerce Protocol](/protocols/agentic-commerce-protocol/) (multi-implementation), [UCP](/protocols/ucp/) (multi-implementation), [AP2](/protocols/ap2/) (multi-implementation), [Trusted Agent Protocol (TAP)](/protocols/tap/) (multi-implementation), [ERC-8183: Agentic Commerce](/protocols/erc-8183/) (shipping) | [Verifiable Intent](/protocols/vi/) (prototype), [ERC-8226](/protocols/erc-8226/) (prototype) |\n| Iframes & server-driven UI | [8 Present](/category/present/) | Text-only chat cannot render interactive approval cards, forms, or state sync safely | [AG-UI](/protocols/ag-ui/) (multi-implementation), [A2UI (Agent to UI) Protocol](/protocols/a2ui/) (multi-implementation), [MCP Apps](/protocols/mcp-apps/) (multi-implementation) | [A2WP](/protocols/a2wp/) (none found) |\n| Syslog, referrer cookies & T&Cs | [9 Account](/category/account/) | Cookies vanish when agents summarize; software processes are not legal persons | [Agent Attribution Protocol](/protocols/agent-attribution-protocol/) (shipping), [Agent Audit Trail](/protocols/agent-audit-trail/) (shipping) | [OpenTelemetry GenAI Semantic Conventions (agents and MCP)](/protocols/otel-genai-semconv/) (prototype), [AIR](/protocols/scitt-ai-agent-execution/) (none found), [Agent Audit Record](/protocols/wimse-agent-audit-record/) (none found), [ADC](/protocols/agent-decision-commitment/) (none found) |\n\n## Contested ground and whitespace by step\n\nSorting the map by running-code maturity (multi-implementation = 2+ independent interoperable implementations; shipping = 1 production implementation; trial / prototype / none-found = non-production or paper spec) shows where the stack has converged, where competing specifications are fighting for the same slot, and where true whitespace remains:\n\n| Step | Total | Multi-implementation (2+ parties) | Single-party shipping | Trial / prototype / spec-only | Contested ground & whitespace implication |\n| --- | --- | --- | --- | --- | --- |\n| [1 Discover](/category/discover/) | 18 | 3 | 6 | 9 | Contested root manifests (llms.txt vs RFC 9727 vs ARD vs SEP-2127); DNS layer still early. |\n| [2 Identify](/category/identify/) | 22 | 5 | 4 | 13 | Converging on RFC 9421 / Web Bot Auth for HTTP signing and ERC-8004 for on-chain registries; 13 drafts competing on workload credentials. |\n| [3 Permit](/category/permit/) | 32 | 6 | 2 | 24 | Most crowded step (32 specs): active standards war in delegated OAuth (ID-JAG, auth.md, AAuth, DAAP) and AI content licensing (Content-Signal vs AIPREF vs RSL). |\n| [4 Read](/category/read/) | 6 | 2 | 3 | 1 | Converging rapidly on Accept: text/markdown, Markdown twins, AGENTS.md, and OKF. |\n| [5 Act](/category/act/) | 13 | 7 | 2 | 4 | Converged around MCP, Agent Skills (SKILL.md), and Zed ACP; in-browser WebMCP in origin trial. |\n| [6 Coordinate](/category/coordinate/) | 13 | 3 | 3 | 7 | Converging on Linux Foundation A2A for HTTP agent cards; pub/sub fan-out and ad-tech negotiation (AAMP) still contested. |\n| [7 Pay](/category/pay/) | 17 | 5 | 4 | 8 | Split into two parallel stacks: HTTP 402 micropayments (x402, MPP, L402) vs card-network checkout mandates (ACP, UCP, AP2, TAP, VI). |\n| [8 Present](/category/present/) | 4 | 3 | 0 | 1 | Small, fast-converging layer (AG-UI, A2UI, MCP Apps); cross-device human interrupt (A2WP) is early whitespace. |\n| [9 Account](/category/account/) | 7 | 0 | 2 | 5 | Largest whitespace on the map (only 1 of 7 specs shipping): post-hoc audit ledgers (SCITT), attribution, and [Agentic EoR](/terms/agentic-eor/) liability wrappers are mostly draft-stage. |\n\n## Harness and client compatibility matrix\n\nA site's published manifests only matter if an agent harness or browser natively parses them. Across the [20 verified harnesses and clients](/instances/#tab-harness) tracked on this site, support currently concentrates on a small core of discovery, read, and tool-invocation standards:\n\n| Harness / Client | Native standards consumed or invoked | Observed |\n| --- | --- | --- |\n| [CopilotKit](https://www.copilotkit.ai) | [AG-UI](/protocols/ag-ui/), [A2UI](/protocols/a2ui/), [MCP Apps](/protocols/mcp-apps/), [A2A](/protocols/a2a/), [MCP](/protocols/mcp/) | 2026-08-14 |\n| [Google Agent Development Kit (ADK)](https://google.github.io/adk-docs/) | [A2A](/protocols/a2a/), [MCP](/protocols/mcp/), [AG-UI](/protocols/ag-ui/) | 2026-07-31 |\n| [Goose (Block)](https://github.com/block/goose) | [MCP](/protocols/mcp/), [AGENTS.md](/protocols/agents-md/), [Agent Skills](/protocols/anthropic-agent-skills-format/) | 2026-08-29 |\n| [GitHub Copilot Coding Agent](https://docs.github.com/en/copilot/how-tos/use-copilot-agents/coding-agent/create-custom-agents) | [AGENTS.md](/protocols/agents-md/), [MCP](/protocols/mcp/) | 2026-08-29 |\n| [pay.sh (Agent Payment CLI and Runtime)](https://pay.sh) | [x402](/protocols/x402/), [MPP](/protocols/mpp/) | 2026-09-04 |\n| [Aider](https://aider.chat/docs/usage/conventions.html) | [AGENTS.md](/protocols/agents-md/) | 2026-08-29 |\n| [Amp](https://ampcode.com) | [AGENTS.md](/protocols/agents-md/) | 2026-08-29 |\n| [Claude Code (Anthropic)](https://platform.claude.com/docs/en/agents-and-tools/claude-code/overview) | [Agent Skills](/protocols/anthropic-agent-skills-format/) | 2026-08-29 |\n| [Cursor](https://cursor.com) | [AGENTS.md](/protocols/agents-md/) | 2026-08-29 |\n| [Devin (Cognition)](https://devin.ai) | [AGENTS.md](/protocols/agents-md/) | 2026-08-29 |\n| [Elemen (AGTP Browser)](https://chrishood.com/introducing-elemen-the-worlds-first-browser-for-the-agent-web/) | [AGTP](/protocols/agtp/) | 2026-06-05 |\n| [Google Chrome (WebMCP Early Preview)](https://developer.chrome.com/blog/webmcp-epp) | [WebMCP](/protocols/webmcp/) | 2026-02-18 |\n\n## Timeline, 2024–2026\n\nThe agentic web's standards did not arrive at once. Jeremy Howard's llms.txt proposal, a plain-Markdown convention for pointing language models at a site's documentation, dates to 2024, with no formal governance behind it. Anthropic published the Model Context Protocol in November 2024. Google followed with the Agent2Agent Protocol in April 2025, now on Linux Foundation track. Coinbase's x402 followed in May 2025, later gaining its own foundation and a 2026 v2 revision. IBM's Agent Communication Protocol, a separate agent-to-agent effort from its BeeAI team, was archived as a standalone specification on 29 August 2025, with its own repository stating it had been folded into A2A. Google launched its Agent Payments Protocol (AP2 v0.1.0) on 16 September 2025 with more than sixty launch partners, and donated the core specification to the FIDO Alliance on 28 April 2026 alongside v0.2.0. Cloudflare published its Agent Skills Discovery RFC (/.well-known/agent-skills/index.json, Draft v0.2.0) in January 2026. Google and industry partners announced Agentic Resource Discovery (ARD) on 17 June 2026, with the specification reaching its v0.91 text on 26 August 2026. The Model Context Protocol's most recent dated specification revision is 2026-07-28, accompanied by SEP-2127 MCP Server Cards (/mcp/server-card, finalized 6 October 2026), RFC 9727 API Catalogs (/.well-known/api-catalog), and Google Cloud's Open Knowledge Format (/okf/index.md). This site re-probes adoption across its /check/ readiness panel daily and publishes the day-over-day findings in the [daily digest](/digest/).\n\n## Colliding acronyms\n\n\"ACP\" is not one protocol; it is four, from four distinct groups, across three different steps of this map. Zed's [Agent Client Protocol](/protocols/agent-client-protocol/) connects code editors to coding agents, at the Act step. AGNTCY's [Agent Connect Protocol](/protocols/agent-connect-protocol/) (Cisco, LangChain, Outshift) defines an OpenAPI execution descriptor and REST runtime lifecycle for invoking remote agents, also at the Act step. IBM's [Agent Communication Protocol](/protocols/agent-communication-protocol/) was an agent-to-agent messaging protocol at the Coordinate step, archived in August 2025 and, by its own maintainers' account, merged into A2A. OpenAI and Stripe's [Agentic Commerce Protocol](/protocols/agentic-commerce-protocol/) lets an agent complete a checkout with a merchant, at the Pay step, and is unrelated to any of the other three. Published guides and summaries routinely state a fact about one of these as though it applied to all four; the full picture is on the [ACP term page](/terms/acp/).\n\n## Frequently confused claims\n\nThree claims recur in vendor and secondary material about this space and do not hold up as stated. Google-Extended does not control Google Search or AI Overviews. It is a separate token from Googlebot, documented as governing use of crawled content by Gemini apps and Vertex AI; Search and AI Overviews are covered by the ordinary Googlebot crawl, so disallowing Google-Extended alone does not remove a site from Google's AI-powered search surfaces. No mainstream answer engine documents reading llms.txt. Google's own generative-search guidance states plainly that no special markup or llms.txt file is required for its AI features, and no first-party documentation from ChatGPT Search, Perplexity, Claude, or Bing/Copilot lists it as something they fetch or use as a ranking or citation input; the file is a real, dated 2024 proposal with real adoption as a navigation aid, but that is a different claim from being consumed by answer engines. \"ACP merged into A2A\" is true only of IBM's Agent Communication Protocol. It says nothing about Zed's Agent Client Protocol, AGNTCY's Agent Connect Protocol, or the OpenAI/Stripe Agentic Commerce Protocol, all three of which remain independent, active specifications; treating the merger as if it settled anything about \"ACP\" in general conflates four unrelated projects that happen to share an initialism.\n\n## Bibliography\n\n- [Model Context Protocol specification, 2026-07-28](https://github.com/modelcontextprotocol/modelcontextprotocol/blob/main/docs/specification/2026-07-28/index.mdx)\n\n- [Agent2Agent Protocol specification](https://a2a-protocol.org/latest/specification/)\n\n- [A2A Agent Card discovery](https://a2a-protocol.org/v1.0.0/topics/agent-discovery/)\n\n- [Agent Client Protocol](https://agentclientprotocol.com)\n\n- [IBM Agent Communication Protocol repository](https://github.com/i-am-bee/acp/)\n\n- [OpenAI/Stripe Agentic Commerce Protocol repository](https://github.com/agentic-commerce-protocol/agentic-commerce-protocol)\n\n- [x402 whitepaper](https://x402.org/wp-content/uploads/sites/10/2026/06/x402-whitepaper.pdf)\n\n- [x402 facilitator documentation](https://docs.x402.org/core-concepts/facilitator)\n\n- [Agent Payments Protocol specification](https://github.com/google-agentic-commerce/AP2/blob/main/docs/ap2/specification.md)\n\n- [AP2 Checkout Mandate](https://ap2-protocol.org/ap2/checkout_mandate/)\n\n- [Agentic Resource Discovery specification](https://github.com/ards-project/ard-spec/blob/main/spec/ard.md)\n\n- [Google's ARD announcement, 2026-06-17](https://developers.googleblog.com/en/announcing-the-agentic-resource-discovery-specification/)\n\n- [RFC 8615, Well-Known Uniform Resource Identifiers](https://www.rfc-editor.org/rfc/rfc8615)\n\n- [OpenAI crawler documentation](https://developers.openai.com/api/docs/bots)\n\n- [Anthropic crawler policy](https://support.claude.com/en/articles/8896518)\n\n- [Perplexity crawler documentation](https://docs.perplexity.ai/guides/bots)\n\n- [llms.txt](https://llmstxt.org)\n\n- [AGENTS.md](https://agents.md)\n\n- [Google's developer guide to AI agent protocols](https://developers.googleblog.com/developers-guide-to-ai-agent-protocols/)\n\n[Markdown](/agentic-web/index.md) · [JSON](/agentic-web/index.json) · [llms-full.txt](/llms-full.txt)\n\n## Frequently asked questions\n\n- **What is the agentic web?**: The agentic web is the ordinary HTTP/DNS web extended with machine-native discovery files, signed identity headers, permission tokens, and invocation or payment protocols that let software agents discover, authenticate, act, and transact without scraping human UI. As of 2026-10-30, agenticweb.wiki maps 132 standards across 9 lifecycle steps (Discover, Identify, Permit, Read, Act, Coordinate, Pay, Present, Account) and 325 verified implementations.\n- **How does the agentic web differ from Agentic AI, the Semantic Web, and Web3?**: Agentic AI is a model capability inside one vendor's runtime, whereas the agentic web is the public protocol surface a site publishes so external agents can interact with it natively. Unlike the Semantic Web, it requires no universal ontology (only narrow, imperative manifests and endpoints), and unlike Web3 it runs on standard HTTP/DNS infrastructure, using blockchain rails only where useful for machine payments (such as x402) or on-chain identity registries (such as ERC-8004).\n- **What are the nine lifecycle steps of the agentic web?**: The nine steps an agent takes across a site are: (1) Discover what the site offers, (2) Identify the agent and origin, (3) Permit or restrict crawler and agent actions, (4) Read machine-formatted content, (5) Act by invoking tools or skills, (6) Coordinate with other agents, (7) Pay or authorize transactions, (8) Present UI to a human, and (9) Account via verifiable audit trails and legal counterparties.\n- **What is the difference between the four ACP protocols?**: ACP refers to four unrelated specifications at different lifecycle steps: Zed's Agent Client Protocol (Step 5, Act — connecting code editors to coding agents), AGNTCY's Agent Connect Protocol (Step 5, Act — OpenAPI execution descriptors and REST runtime lifecycle), IBM's archived Agent Communication Protocol (Step 6, Coordinate — merged into A2A in August 2025), and the OpenAI/Stripe Agentic Commerce Protocol (Step 7, Pay — merchant checkout for AI agents).\n"
}