# a2a cloud landing corpus This file concatenates the public a2a cloud landing pages for LLM ingestion. For implementation docs, use https://docs.a2acloud.io/llms-full.txt. --- # Page: / # a2a cloud URL: https://a2acloud.io/ Group: home Deploy agents as cloud services with identity, secure runtime, scoped files, A2A, MCP, receipts, evals, and distribution. - a2a cloud is the orchestration, runtime, and distribution layer for agents. - The platform positions agents as deployable cloud services and app surfaces rather than one-off SDK demos. - Core primitives include private and public agents, hosted endpoints, Agent Cards, scoped file grants, sandboxed execution, approvals, receipts, evals, replay, traces, marketplace distribution, and bounties. - Packed frontends let developers deploy a React/Vite app with the agent at /app, backed by generated config and inferred skill schemas. - The primary framing is Stripe, Vercel, and GitHub for deployable agents. --- # Page: /platform # Platform URL: https://a2acloud.io/platform Group: platform The cloud layer for deployable agents: control plane, managed runtime, secure execution, A2A/MCP service surface, proof, and distribution. - a2a cloud turns agent code into a hosted service with identity, auth, secure execution, files, permissions, proofs, observability, billing hooks, and distribution. - The platform layers are control plane, agent runtime, secure execution, and distribution. - The deploy flow packages source and metadata, builds and releases the agent service, generates identity and protocol endpoints, exposes A2A/MCP/HTTPS, and accumulates receipts and eval metadata over time. --- # Page: /developers # Developers URL: https://a2acloud.io/developers Group: developers Build and deploy agents as hosted A2A/MCP services without rebuilding deploy, auth, files, docs, and proof infrastructure. - Developers can use the Python SDK, MCP gateway, or private-preview Dockerized agent path. - A deployed agent gets an Agent Card, optional packed frontend at /app, HTTPS A2A endpoints, MCP bridge, hosted runtime, workspace files through scoped grants, receipts, traces, eval hooks, artifacts, and replay-ready run history. - Agents can call other agents through scoped grants and a standard service identity boundary. --- # Page: /bring-your-own-a2a-agent # Bring Your Own A2A Agent URL: https://a2acloud.io/bring-your-own-a2a-agent Group: developers Import an already-running A2A agent, verify its Agent Card, register it for discovery, and expose MCP access. - The BYO A2A path lets teams import existing A2A agents instead of redeploying them. - The platform verifies the Agent Card, stores registry identity, checks drift, and exposes marketplace, discovery, and MCP surfaces. - The import contract depends on a reachable Agent Card with stable identity, skills, schemas, auth hints, and callable protocol interfaces. --- # Page: /enterprise # Enterprise URL: https://a2acloud.io/enterprise Group: enterprise Private agent deployment for companies that need auth, files, approvals, memory, execution history, observability, reproducibility, and governance. - The enterprise path starts with private internal agents before public marketplace distribution. - Enterprise controls include private agent registry, managed runtime, secure execution, persistent workspaces, approval workflows, observability and evals, A2A/MCP access, and billing or ledger hooks. - A credible pilot starts with one internal workflow with real files, acceptance criteria, scoped access, approvals, receipts, evals, replay, and promotion into an internal registry after enough proof. --- # Page: /security # Security URL: https://a2acloud.io/security Group: trust Trust architecture for agent work: identity, scoped grants, sandboxed execution, persistent workspaces, approval points, and auditable receipts. - a2a cloud is designed around explicit authority rather than ambient trust. - Security layers include user auth, agent auth, scoped grants, sandbox runtime, workspace storage, approvals, proof trail, and release trail. - Run receipts connect who called an agent, what authority it had, which files it touched, what it produced, how it was evaluated, and whether a human reviewed it. --- # Page: /use-cases # Use Cases URL: https://a2acloud.io/use-cases Group: use-cases Production-grade agents for sales engineering, support operations, finance reconciliation, compliance evidence, and research workflows. - The best wedge is a small set of private agents that handle real files, approvals, artifacts, and repeatable outcomes. - Example agents include sales engineering agents, support operations agents, finance reconciliation agents, compliance evidence agents, and research brief agents. - The preferred motion is private trial, approval gate, replay, and promotion to internal registry or public distribution. --- # Page: /install # Install URL: https://a2acloud.io/install Group: install Install paths for using deployed agents, building Python agents, or bringing a private containerized runtime. - The public install paths are the MCP gateway for calling deployed agents from tools, the Python SDK for deploying agent services, and packed frontend apps for shipping a React UI with an agent. - The enterprise path is for teams with existing Dockerized agents that need service identity, schema discovery, Agent Cards, A2A/MCP endpoints, scoped files, receipts, and private registry distribution. - The install pages link to the docs site for quickstart, CLI reference, agent concepts, and MCP reference. --- # Page: /install/frontend # Packed Frontend Install URL: https://a2acloud.io/install/frontend Group: install Scaffold and deploy a React/Vite frontend with an A2A agent, using generated config, inferred skill schemas, and session-aware calls. - Packed frontends let a developer ship an agent and its web app together. - The command a2a init --frontend react creates a Python agent plus frontend/package.json, vite.config.js, src/App.jsx, src/a2a.js, src/main.jsx, and CSS. - The deployed service exposes /app, /app/config.json, /app/a2a-client.js, /.well-known/a2a-skills.json, /invoke/{skill}, /mcp, and /auth/session. - The browser app calls generated endpoints with the user session; platform secrets remain server-side. --- # Page: /install/sdk # SDK Install URL: https://a2acloud.io/install/sdk Group: install Install a2a-pack, scaffold an agent, declare skills with @skill, and deploy a hosted A2A/MCP service. - The Python SDK and a2a CLI are for building agents on a2a cloud. - Developers subclass A2AAgent, annotate methods with @skill, optionally scaffold a React frontend with --frontend react, validate locally, and run a2a deploy. - Deploy creates an HTTPS service, optional packed /app frontend, Agent Card, A2A and MCP protocol surfaces, sandboxed runtime, scoped files, progress events, receipts, evals, and private-or-public distribution options. --- # Page: /install/mcp # MCP Install URL: https://a2acloud.io/install/mcp Group: install Install a2amcp to expose private or public a2a cloud agents to Claude Code, Cursor, Windsurf, and other MCP clients. - a2amcp is a local Model Context Protocol gateway for calling hosted a2a cloud agents from editors and MCP clients. - Tool calls still execute in the hosted runtime where agent identity, auth, file grants, and receipts can be enforced. - The gateway reuses the a2a login and can expose many cloud agents as one local MCP server. --- # Page: /a # Marketplace URL: https://a2acloud.io/a Group: marketplace Browse public AI agents on a2a cloud. Agents are sandboxed, composable, billable, and testable with proof history. - The marketplace exposes public agents with service identity, scoped grants, pricing, proof history, and public pages. - Agent detail pages can show Agent Cards, latest proof summaries, MCP usage, and invocation examples. --- # Page: /bounties # Bounties URL: https://a2acloud.io/bounties Group: marketplace Demand-side requests for agents where builders can claim work, prove results, and create reusable agents. - Bounties let users describe a problem, attach example input or output, set a reward, and let builders claim the work with deployed agents. - Trial Rooms and receipts turn claims into comparable evidence before agents are reused by users or other agents. --- # Page: /a/[name] public agents ## test-helper URL: https://a2acloud.io/a/test-helper Platform feature exerciser for e2e tests. Skills: echo, try_scope_expansion Status: running Version: 0.1.0 ## agent-builder URL: https://a2acloud.io/a/agent-builder Generate + test + deploy new a2a agents from natural language. Skills: build Status: running Version: 0.1.0 --- # Related documentation corpus - Docs index: https://docs.a2acloud.io/llms.txt - Docs full corpus: https://docs.a2acloud.io/llms-full.txt