Skip to guide
IndigoIndigo · HQ Field Guide
Chapter 13 / 15Contents
Chapter 13shared · 13 min

Checkpoints, handoffs, and learning

Keep context fresh, move work between sessions, and turn corrections into institutional memory.

01Use checkpoints and handoffs at the right moments.
02Capture durable corrections as learning instead of repeating them in every session.
$ /checkpoint
Fig 13.0 · Durable memory
The concept

Durable memory

Checkpoints preserve state, handoffs transfer work, and learnings improve future behavior.

DURABLE MEMORY ACROSS SESSIONSSESSION ASESSION BCHECKPOINTCONTEXT FRESHENDHANDOFFSTATE.JSONRESUMERULELEARN — CORRECTION BECOMES A RULEPERSISTS BELOW THE TIMELINE · REAPPLIED EVERY SESSION
Fig 13.1 · Checkpoint pauses
Step 01Save point

Checkpoint pauses

A checkpoint is lightweight session memory for moments when you want to save without stopping.

Checkpoint after meaningful file edits, test runs, commits, useful discoveries, or when pausing a thread.

The next session can resume from the checkpoint without a long recap.
/checkpoint/startwork
Step 02Transfer

Handoff transfers

A handoff is a deliberate transfer package with current progress, remaining work, blockers, and state.

Run handoff when ending a work arc, switching sessions, preparing someone else to continue, or before launching a major /run-project task.

A fresh session can pick up the work without re-investigating from scratch.
/handoff/startwork
Fig 13.2 · Handoff transfers
Fig 13.3 · Learning persists
Step 03Rule card

Learning persists

Learning turns user corrections, team preferences, and repeated patterns into institutional memory.

Capture rules as specific if-then guidance: when this situation occurs, do this instead. Avoid vague instructions like be thorough.

Future sessions apply the lesson without needing the same correction.
/learn/checkpoint/handoff
Commands

Keep the chapter executable.

Checkpoint
$ /checkpoint

Writes lightweight state so the current work can resume later.

Handoff
$ /handoff

Prepares a new session to continue the work from a clear state.

Learn
$ /learn

Captures durable corrections, policies, or insights for future sessions.

References

What to keep in mind.

When to checkpoint

After meaningful edits, test runs, commits, or any pause where state would be costly to reconstruct.

When to learn

After a user correction, repeated failure mode, or reusable operating rule.

Next
Next move

Now learn how to recover when setup or sync does not look right.

Continue to troubleshooting
Part 13 · Checkpoints, handoffs, and learningIndigo · HQ Field Guide