INDEPENDENT TRUST LAYER FOR AI AGENTS
OpenAgentSource is an independent trust registry for AI agents — a place to discover, inspect, review, and verify portable AI agents before you use them.
See who built an agent, what it can access, how it runs, and whether it has been reviewed before you install or use it.
For builders: turn your GitHub README into a public trust profile your users can review before installing your agent.
by Northstar Labs · agent.open/v1.4.2
Trust History
Stable across 4 releasesLast reviewed 3 days agoFor agent builders
OpenAgentSource turns your agent’s README, manifest, source links, install details, permissions, and risk disclosures into a public trust profile. You get a profile users can review, a README badge, and a launch kit you can share.
A clean page with source links, install details, permissions, risk rationale, and creator information.
Add a trust badge to your GitHub README so users can review your profile before installing.
Copy-ready LinkedIn, short post, README block, and email copy after approval.
Review your draft profile before it goes public.
OpenAgentSource documents risk, permissions, and external services in a format users can understand.
Approved Registry Activity
Recently approved agent listings with visible creator, risk, manifest, and validation signals.
Each profile documents source links, install instructions, permissions, external services, risk rationale, and creator details.
NexusPay is a reference implementation of an autonomous, budget-aware AI agent that buys its own data over the x402 HTTP payment protocol. You ask a question; it plans which paid sources to buy (Gemini), enforces per-query & daily budget caps, settles each purchase with testnet USDC over x402, and synthesizes everything it bought into one answer.
Describe a landing page in plain words — your AI builds it, validates it, and ships it straight to WebCake. No drag-and-drop, no JSON.
Astakos is a local-first AI companion and multi-agent personal assistant that learns routines, remembers long-term context, and helps proactively across Telegram, Web UI, and CLI.
Creates a short revenue opportunity brief from CRM and meeting notes.
Flags lease clauses that may need legal, financial, or operational review before signing.
Suggests objection-handling talking points grounded in product notes and approved sales messaging.
Know a builder creating serious agents? Invite them to submit a trust profile.
01 / Brand Foundations
OpenAgentSource is a trust layer, not an agent runtime. We do not execute agents or control third-party agents. We help users review manifests, permissions, risk labels, creator identity, and trust history before use.
02 / Trust Language
03 / Product Application
Agent card visual layout
Research Brief AgentTurns long-form market reports into structured summaries with source notes and risk labels.
Agent detail page concept
Trust summary above the foldValidation, creator, permissions, manifest YAML, health checks, and install instructions live where buyers expect to inspect them.
Creator profile concept
Reputation details, verification status, specialty, website, GitHub, and listed manifests.
Verified CreatorFeedback loop concept
Help shape the trust layer.Early buyers and creators can tell us what feels useful, confusing, risky, missing, or worth paying for.
04 / Launch and Handoff
Join the private beta as a buyer or creator, or send feedback about the trust signals you need before evaluating portable agents.
