Examples Library
AgentXchain ships two different kinds of examples:
- governed product examples that prove AgentXchain can take a real product from idea to runnable software
- proof and connector examples that prove runner, MCP, remote-agent, and remote-verification boundaries honestly
This page is the hub. Each example below has its own dedicated docs page under /docs/examples/*.
Every example page explains:
- what the example is
- the role/workflow shape or proof surface it uses
- how AgentXchain governs the build or enforces the contract
- how to run it
- key takeaways
Governed Product Examples
| Example | Category | Why It Matters |
|---|---|---|
| Habit Board | Consumer SaaS | Designer-in-the-loop product workflow with a dedicated design phase |
| Trail Meals Mobile | Mobile | React Native / Expo governance with offline-first architecture |
| Async Standup Bot | B2B SaaS | Integration and operations phases as first-class governed work |
| Decision Log Linter | Developer Tool | CLI product with architecture and release phases |
| Schema Guard | Open Source Library | API review and release engineering for an npm library |
| Governed Todo App | Tutorial App | Minimal PM -> Dev -> QA lifecycle walkthrough |
Runner And Live Proof Examples
| Example | Surface | Why It Matters |
|---|---|---|
| CI Runner Proof | runner-interface | Repo-native proof for non-CLI runner authors |
| External Runner Starter | installed package starter | Canonical starting point for external runner authors |
| Live Governed Proof | live provider proof | Real-model governed execution across shipped runtime paths |
| Checkpoint Handoff Proof | auto-checkpoint proof | Evidence that continuous writable handoffs leave checkpoint commits and events after every accepted turn |
| Live Continuous 3-Run Proof | continuous case study | Public evidence for three unattended governed runs with real Anthropic QA review |
| Live Dogfood Proof: tusq.dev | dogfood proof | Three full governed runs on a real production codebase with perpetual idle-expansion |
MCP Examples
| Example | Transport | Why It Matters |
|---|---|---|
| MCP Echo Agent | stdio | Smallest possible governed MCP tool contract |
| MCP HTTP Echo Agent | streamable HTTP | Remote MCP transport proof |
| MCP Anthropic Agent | stdio + Anthropic | Real-model MCP execution |
Remote Execution Examples
| Example | Surface | Why It Matters |
|---|---|---|
| Remote Agent Bridge | remote_agent | Proposal/apply and review-only remote execution |
| Remote Conformance Server | verify protocol --remote | Runnable remote verifier contract |
Provenance And Governance
The governed product examples are not stubs. They include real code, tests, workflow-kit artifacts, TALK.md provenance, and governed configs with explicit roles and gates.
The proof examples are intentionally different: they do not pretend to be full product apps. Their job is to prove the runner, MCP, remote-agent, or verification boundary honestly.
Start Here
- If you want the cleanest product walkthrough, start with Governed Todo App.
- If you want the strongest product proof, read Habit Board and Schema Guard.
- If you are integrating AgentXchain into another runtime, read CI Runner Proof, External Runner Starter, and the MCP examples.
- If you need live runtime proof, read Live Governed Proof, the dated Live Continuous 3-Run Proof, and the Live Dogfood Proof.