majiayu000 update-todos
A task list drifts the moment work starts. Items get done sideways, get overtaken by a design change, or turn out to be blocked on something nobody saw. A list that no longer matches the tree is worse than no list, because it is trusted.
This skill is temporal: it reconciles the list against what actually happened. For a list that is accurate but too coarse to execute, use sophisticate-todos instead.
Reconcile
Three-way comparison: the list as written, the tree as it stands, and what the conversation established.
Classify every existing item exactly once:
| Class | Meaning |
|---|---|
landed |
Done, with proof |
still-open |
Unchanged, still required |
overtaken |
A design change made it unnecessary |
blocked |
Cannot proceed until something external clears |
newly-discovered |
Not on the list; found during the work |
A landed claim requires proof: the test, the command output, or the path:line that demonstrates it. An unproven completion stays still-open. Someone saying an item is done is not proof; it is the claim under test.
Completion criterion: every pre-existing item carries exactly one classification, and every landed cites its proof.
Apply
Write the reconciled list back through the todo tool.
An overtaken item is dropped with a one-line reason recorded in the report, never deleted silently. A dropped item with no recorded reason is indistinguishable from an item you forgot.
Completion criterion: the written list contains no item whose classification contradicts its recorded state.
Report
Emit the delta only: what changed classification, and why.
Completion criterion: the report accounts for every item whose state moved, and says nothing about items that did not move.