What We Need
DRS has a working verifier, a TypeScript SDK, and a spec. What it does not have is a body of real examples showing it plugged into actual agentic systems doing actual work.
This issue is an open invitation. Pick a system you already use or maintain. Wire DRS into it. Show what it looks like.
The One Rule
Do not create a new toy project just for this. That produces examples nobody learns from because the context is fake. Use something real — a LangChain agent, an AutoGen workflow, a custom MCP server, a CrewAI crew, an OpenAI function-calling loop, a home automation bot, a CLI tool that calls external APIs. If it delegates actions on behalf of a human, it qualifies.
What Makes a Good Example
- The agent actually does something (calls a tool, reads a file, sends a request)
- The delegation receipt is issued before the action and verified before execution
- A human consent record exists where a human is in the loop
- The audit trail is readable — you can look at the receipts and understand what happened and who authorized it
- It runs. Not "here is the concept" — the code runs.
What We Are Not Looking For
- Architecture diagrams with no code
- Examples that mock the verifier
- Wrappers around the SDK that do not actually invoke any tool
- "Coming soon" placeholders
Format
Put your example in examples/ at the root of this repo, in its own directory with a short name that describes the system (examples/langchain-web-agent/, examples/mcp-file-server/, etc.).
Include a README.md that explains:
- What the agent does
- Where in the code DRS verification happens
- How to run it (dependencies, env vars, one command to start)
That is the whole requirement. No template to fill out. No process to follow. Just working code with a readable README.
Why This Matters
The spec is a spec. The verifier verifies. But until someone can point at a real system and say "this is what DRS looks like in production," it is an abstraction. Examples are how people decide whether a piece of infrastructure is worth adopting.
If you have questions about how to wire DRS into a specific system, open a comment here. If you get stuck on the SDK or verifier, open a separate issue with the specific problem.
What We Need
DRS has a working verifier, a TypeScript SDK, and a spec. What it does not have is a body of real examples showing it plugged into actual agentic systems doing actual work.
This issue is an open invitation. Pick a system you already use or maintain. Wire DRS into it. Show what it looks like.
The One Rule
Do not create a new toy project just for this. That produces examples nobody learns from because the context is fake. Use something real — a LangChain agent, an AutoGen workflow, a custom MCP server, a CrewAI crew, an OpenAI function-calling loop, a home automation bot, a CLI tool that calls external APIs. If it delegates actions on behalf of a human, it qualifies.
What Makes a Good Example
What We Are Not Looking For
Format
Put your example in
examples/at the root of this repo, in its own directory with a short name that describes the system (examples/langchain-web-agent/,examples/mcp-file-server/, etc.).Include a
README.mdthat explains:That is the whole requirement. No template to fill out. No process to follow. Just working code with a readable README.
Why This Matters
The spec is a spec. The verifier verifies. But until someone can point at a real system and say "this is what DRS looks like in production," it is an abstraction. Examples are how people decide whether a piece of infrastructure is worth adopting.
If you have questions about how to wire DRS into a specific system, open a comment here. If you get stuck on the SDK or verifier, open a separate issue with the specific problem.