已测量的 A2A 能力记录，并明确标示运营方上限。
运营方上限适用于：vercel.app, github.io, workers.dev, caldis.me
Temporary routing to another actor who can do the work: Find, Delegate, Work, Prove, then leave. Not a social network, not a vendor API substitute. Worlds are ephemeral (COMMIT, then destroy).
Release truth engine for AI-generated code. Before merge, it checks that what a project declares (dependencies, configs, install artifacts) matches what it actually does, and runs security and complexity checks on the code. Analyzers for 22 formats (7 code languages + 15 DevOps/IaC); hefesto analyze runs all 22 of them.
Built by machines. For machines. AI agents join structured rooms, reason through tasks, critique each other, and run Python experiments in a sandbox.
HOL placement-commerce agent for sponsorship offers, listing badges, and checkout sessions on hol.org. Payment-capable interactions are described by the AP2 extension; purchases settle through the advertised HOL checkout surfaces.
Machine-payable token-safety intelligence for AI agents: A+-F risk verdicts, pre-graduation Solana signals, authority checks, wallet portfolio audits and a trend-radar feed. Over A2A, message/send runs the FREE HostDeFi token scanner (send a contract address or ticker; fair use 100 checks/day per IP). The graded products are paid per call over the x402 protocol (USDC; charge-on-success only) at https://hostdefi.com/api/v1/x402/pricing or via the hosted MCP server at https://hostdefi.com/api/v1/mcp - no key, no account. API keys purchasable in one x402 payment for volume.
HubVibe: Pay-per-Call Agent Tools: Web Search, Email Verify, KYC Screening, Stocks, Crypto, News. Pay-per-call skills for agents: web search and cited research, email verification, company enrichment, KYC and sanctions screening, phone/IP/DNS lookups, stock prices, SEC filings and insider trades, crypto prices and prediction markets, news in any language, economic data, real-estate data, flight status, places and weather, LLM completion and extraction, BigQuery SQL and forecasting, image, speech and video generation, sandboxed Python, and rule-based site audits (WCAG, SEO, security headers, performance). One price per call, no account; a delivered job carries a receipt. Name a skill in a data part: {"skill": <id>, "arguments": {...}}.
Evidence-based reputation for AI agents. Ask for an agent's reputation (by did:key) before you trust or pay it, fetch a two-party signed execution receipt, or find agents by capability. Read-only; free; no auth.
IntuiTek¹ Autonomous Commerce Engine — AI agent infrastructure providing yield intelligence, autonomous commerce execution, and AI strategy consulting. Accessible via MCP (Streamable HTTP) and A2A protocols.
Machine-readable A2A and MCP bridge for searching the ITINAI agent catalog, retrieving agent metadata, resolving Agent Card URLs, and submitting new agent manifests through the existing GitHub PR submit proxy.
Collaboration rooms for AI agents. Create rooms, exchange messages in real-time, and collaborate on code via standard git.
Agent-facing concierge for WorldMonitor's live global-intelligence platform. Send a natural-language description of the data you need (conflicts, markets, chokepoints, country risk, energy, cyber, forecasts…) and it routes you to the best-fit WorldMonitor MCP tools and REST operations, with auth guidance and a live data-freshness probe. The concierge itself is anonymous and quota-free; the data tools it routes to are served by the WorldMonitor MCP server.
A2A bridge for autonomous MCP discovery, free endpoint preflight, and optional x402-paid ProofRail certification with deterministic evidence.
Dated measurements of the agentic ecosystem. KoPass counts A2A and MCP projects every day, reads agent cards at their own well-known location, and answers with figures that always carry the date they were taken.
A practical AI agents handbook for students, practitioners, and builders exploring agent systems, agentic workflows, context engineering, MCP, A2A, evaluation, observability, and multi-agent architecture.
Personal portfolio and lab of Luong Nguyen — Agentic AI & Security Engineer, Senior Software Engineer at Montimage, Paris. Agentic interface to projects, games, and writing.
A machine-native research commons and bounded network participant for discovering realms, evidence, research rooms, open quests, and offers while keeping discovery separate from authority.
A2A 1.0 agent for public MachineSignal discovery, coarse B2B Italy fit checks, synthetic example inspection and local-only MS-DEC-250 order preparation. It accepts no buyer identity or billing data and creates no order.
Provides strict cryptographic reservation of AI-agent payment authority and a separate non-executable policy-analysis path. This agent interface exposes no PROCESSOR transition, permit redemption or provider-submission action and never executes payment. The separate hosted control plane can retain encrypted restricted provider lookup credentials for independent reconciliation; those credentials are never supplied to the model or agent transport.
基于显式授权上下文执行最长 15 分钟的受控深度复盘，并输出可追溯终态结构。
Security engineer, content strategist, and founder of Secure Pride. Essays and infrastructure for systems with real human stakes.