What you’ll get from this lesson: Create a handoff card so a new AI first checks the current information, explains its understanding, and then acts within its authorized scope.
Prerequisites: No software installation required. Prepare a work example that can be shared publicly or has had sensitive information removed.
In the original recording, I compared a new AI to a colleague who has just joined the company. To help it understand what is currently being worked on, it needs documents and work-status information it can look up. Explaining everything again each time can leave the latest progress scattered across different conversations. I want whoever takes over to return to the same source to check.
Start with the Goal It Needs to Complete Now
The first item in the handoff information is what needs to be accomplished this time and what counts as complete. Next, write down how far the work has progressed and what remains. This information needs a date or an identifiable version so that past to-do items are not mistaken for the current state.
Even without a task board, you can start with a single document. The point is to leave a reference that can be checked again, not to build a complete ticketing system first.
Include Both Work Rules and Access Points for Tools
The information should clearly state what is allowed this time, what is not allowed, and which document or tool needs to be used. The original draft used the development console and a spreadsheet-organizing tool as examples. This lesson takes only the handoff concept from those examples; it does not require beginners to build an advanced task board first.
If you later provide tools through MCP, expose only the scope needed for the current task. Being able to check work progress and being allowed to modify all work items are different permissions.
Have It Report Its Understanding Before Deciding Whether to Act
Ask the new AI to read the specified materials first, then list the current goal, progress so far, next steps, and anything that remains unconfirmed. Each item should identify which source it came from. If the version it describes does not match the current work, clarify that before rushing to begin.
This handoff card is a practice tool developed from the original draft’s “new colleague reading the manual” analogy. It does not replace the complete development-communication course taught personally by Darks.
Task Status Must Be Traceable to Evidence of the Results
Seeing only three columns—To Do, In Progress, and Done—is not enough to determine what was actually accomplished. For a fictional task such as “organize work notes,” completion should at least identify the document produced and how you checked that its contents were correct.
Even when AI says a task is complete, you still need to review the result. If there is no source or it has not been tested, keep its status unconfirmed so that the next person or AI taking over does not rely on an unreliable conclusion.
Today, Use One Card to Simulate a Handoff
Put the goal, progress, rules, and access points for the materials on the same card. Start a fresh conversation, provide only the materials that are permitted to be used, and ask the AI to begin with a read-only summary. Compare its account of the goal and status with your information, and check whether it has missed any restrictions.
When a new AI can return to the same supporting information, it becomes easier for it to pick up your work. You also begin turning your ways of working into assets you can continue to maintain.
These seven articles are preparation for understanding your options and making choices. If your environment is not set up yet, return to the course map and complete L0’s environment check. Once you are ready, check the current release status of L1’s free hands-on lessons.
👉 Start L1-01: build your first Agent →
L1 is free after you sign in. Start with L1-01, follow it step by step, and you'll have an Agent of your own.
Not ready to start yet? That's fine. Scroll to the bottom and tap "Next lesson →" to keep reading the free articles, then come back to L1-01 when you're ready.