# Agent Interface > agentinterface.app is the reference site for agent interfaces — the protocols that connect AI agents to software (MCP and its peers), and the interface patterns that keep humans in command of agents (approvals, streaming progress, handoffs). It maintains a continuously updated tracker of the protocol landscape and publishes working guides. ## Core pages - [What is an agent interface?](https://agentinterface.app/): the definitional hub — the term, the two layers (human-agent and agent-software), the protocol map, and FAQs - [The agent-interface tracker](https://agentinterface.app/tracker): a living, dated index of agent-interface protocols, patterns, and surfaces, with status calls - [Guides](https://agentinterface.app/guides): working guides to protocols and agent-UX patterns ## For agents This site speaks MCP. The tracker is available as callable tools, not just as a page — add `https://agentinterface.app/mcp` as a custom connector (Streamable HTTP, JSON-RPC over POST) in Claude, ChatGPT, Perplexity, Grok or Mistral. Tools: `list_agent_protocols`, `get_agent_protocol`, `search_agent_protocols`. Read-only; facts are facts and the editorial calls are CC BY 4.0 with attribution to agentinterface.app. The same data is at `https://agentinterface.app/tracker.json`. ## Guides - [OpenAI's app directory quietly became a plugin standard](https://agentinterface.app/guides/openais-app-directory-quietly-became-a-plugin-standard): In ten months, OpenAI's ChatGPT App Directory went from a proprietary store to a directory built on a cross-vendor plugin spec it helped write. - [Multi-agent supervision: the fleet view arrives](https://agentinterface.app/guides/multi-agent-supervision-the-fleet-view-arrives): Claude Code, Devin Desktop, and Cursor all shipped a dashboard for watching several agents at once this year — and independently landed on the same shape. - [MCP Apps: UI enters the protocol, officially](https://agentinterface.app/guides/mcp-apps-ui-enters-the-protocol-officially): Tool UI went from proposal to formal MCP extension in eight months. The mechanics, the client list, and what security researchers are already flagging. - [Agent Plugins: a shared format for skills and MCP](https://agentinterface.app/guides/agent-plugins-a-shared-format-for-skills-and-mcp): Six companies agreed on one plugin format for AI agent skills and MCP servers. Anthropic, whose specs it packages, isn't among them. - [OS-level agent workspaces: the OS as the permission gate](https://agentinterface.app/guides/os-level-agent-workspaces-the-os-as-the-permission-gate): Windows 11 and Apple are both moving agent permissions into the OS and the app-API layer. Here's how the two mechanisms actually differ. - [ACP one year in: the LSP for coding agents](https://agentinterface.app/guides/acp-one-year-in-the-lsp-for-coding-agents): Zed's Agent Client Protocol went from a Gemini CLI side project to JetBrains' default and a public agent registry — with v2 now in draft. - [Cost transparency: what to show before an agent spends](https://agentinterface.app/guides/cost-transparency-what-to-show-before-an-agent-spends): Token spend and real-money spend are tracked by completely different systems today, and neither shows you the number before you say go. - [The empty prompt problem: what to show first](https://agentinterface.app/guides/the-empty-prompt-problem-what-to-show-first): A blank text box asks new users to recall what an agent can do before they have seen it work. Here is what gets people to a first real result instead. - [llms.txt in 2026: adoption is up, readership still isn't](https://agentinterface.app/guides/llms-txt-in-2026-adoption-up-readership-still-isnt): More sites ship llms.txt than ever, but AI crawlers still ignore it — while Chrome's own Lighthouse tool just started auditing for one. - [Trust calibration: why agents should show their doubt](https://agentinterface.app/guides/trust-calibration-why-agents-should-show-their-doubt): Agents sound equally confident whether they're right or guessing. Calibrated uncertainty is a design problem, not just a training one. - [Streaming progress: what to show while an agent works](https://agentinterface.app/guides/streaming-progress-what-to-show-while-an-agent-works): Not every tool call earns a line on screen. What shipped agent products actually stream, what they hold back, and where the line falls. - [Agentic browsers in 2026: what actually shipped](https://agentinterface.app/guides/agentic-browsers-in-2026-what-actually-shipped): Atlas dies August 9, Comet went free and cross-platform, and Edge and Brave folded agentic browsing into browsers people already use. - [Computer use in 2026: three different bets on the screen](https://agentinterface.app/guides/computer-use-in-2026-three-different-bets-on-the-screen): Anthropic, OpenAI, and Google each shipped computer-use agents, and disagree on whether the right scope is the desktop, the app, or the browser. - [Handoff design: escalation without losing context](https://agentinterface.app/guides/handoff-design-escalation-without-losing-context): When an agent hands off to a human, the failure mode isn't the escalation — it's the context that doesn't survive it. - [AP2 and x402: two answers to how agents pay](https://agentinterface.app/guides/ap2-and-x402-two-answers-to-how-agents-pay): Google's mandate-based AP2 and Coinbase's HTTP-native x402 solve different halves of agent commerce, and increasingly plug into each other. - [MCP goes stateless: what the 2026-07-28 spec changes](https://agentinterface.app/guides/mcp-goes-stateless-2026-07-28-spec-changes): MCP's biggest spec revision yet ships today: sessions are gone, replaced by explicit state. Nothing breaks immediately — but plan the migration. - [When a long-running agent should ping you](https://agentinterface.app/guides/when-a-long-running-agent-should-ping-you): A three-tier framework for agent notifications, presence-aware suppression, and what separates a useful ping from noise. - [AG-UI in 2026: fast adoption, no foundation yet](https://agentinterface.app/guides/ag-ui-in-2026-fast-adoption-no-foundation-yet): CopilotKit's frontend protocol landed in AWS Bedrock and raised $27M this year, but it's still the one major agent protocol without a foundation home. - [Undo for agents: reversibility as the safety net](https://agentinterface.app/guides/undo-for-agents-reversibility-as-the-safety-net): Approval gates ask permission before an action. Undo is the backstop for what happens after — and coding agents are quietly building it out. - [What to log so an agent's audit trail survives an incident](https://agentinterface.app/guides/what-to-log-so-an-agents-audit-trail-survives-an-incident): Agent audit trails need more than request logs — they need reasoning, authorization, and identity. What regulators and cloud vendors now require. - [Stopping an Agent Mid-Task Without Corrupting State](https://agentinterface.app/guides/stopping-an-agent-mid-task-without-corrupting-state): Every serious agent surface needs a stop button. Most don't have one that reliably works, and the ones that do show why it's hard. - [Who Approves What: Enterprise Agent Governance in 2026](https://agentinterface.app/guides/enterprise-agent-governance-who-approves-what-in-2026): Agent identity platforms from Microsoft and Okta went GA this year. Adoption is still outrunning the controls meant to govern it. - [The agent graveyard: what died in two years](https://agentinterface.app/guides/the-agent-graveyard): Operator, Atlas, Agent Builder, IBM's ACP — flagship agent surfaces that died in months, and the three patterns in how they went. - [MCP vs A2A: which layer each protocol owns](https://agentinterface.app/guides/mcp-vs-a2a): They aren't competitors — they're adjacent layers. Where the seam sits, how both races ended, and when your architecture needs each one. - [Permission modes: graduated autonomy for AI agents](https://agentinterface.app/guides/permission-modes-graduated-autonomy-for-ai-agents): The four-rung autonomy ladder shipped agent products converged on, and the rules that make it earn trust instead of burning it. - [The approval gate: sign-off that stays meaningful](https://agentinterface.app/guides/the-approval-gate-designing-sign-off-that-stays-meaningful): Why agent approval gates decay into click-throughs, and the five design moves that keep human sign-off worth something.