Write a structured checkpoint ("handoff") that lets the next session resume cleanly — what changed, where things stand, the next concrete action, system state, and files touched. Invoke before a context limit, at session end, when the user says "checkpoint", "handoff", "save state", or "wrap up", and proactively when a long session is nearing its limit.
SkillScore100Maintained
Automatic
◆Knowledge Management pack
Last verified 9/19/2026
What it changes
Passed 20 of 20 test cases with the skill, and 5 of 20 without it.
Without skill23%
With skill100%
Checks it passed that the model alone missed
The seat field is written as VP4.
The sequence count displays as 100.000.
The sequence field is padded as 12.008.
and 20 more in the full report
17 fixed5 passed either way0 still failing0 broke
Pass-rate lift
+77 pp
Tokens
-44%
Agent turns
0%
Response time
-42%
Model gemini-3.6-flashMeasured 9/19/2026v222 of 22 cases passed with skill · 20 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 15 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 15
Prompt
Please prepare a handoff entry for node DEV-04 operating on seat VP2 with master epoch 15 and local count 3. Add this checkpoint line to the top of our repo's handoff file on main.
What was checked, without and with the skill
Checked
Without skill
With skill
The sequence field in the checkpoint line uses 15.003.
Missed
Passed
The target file path is handoff/LATEST_HANDOFF.txt.
Missed
Passed
case-02 · gemini-3.6-flash · 9/19/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.