refactor: prune core tool and prompt surface
This commit is contained in:
21 files changed
+50
-139
No files matched your search
@@ -899,12 +899,8 @@ async function runStep(
|
||||
})
|
||||
session.append('tool/result', {
|
||||
turn, step,
|
||||
// The correlation id MUST be the loop's authoritative call.id (the
|
||||
// model-transcript id that deriveMessages turns into toolCallId), NOT
|
||||
// result.callId — a post-execute waterfall listener returning a
|
||||
// mismatched id would otherwise orphan the call↔result pairing in the
|
||||
// next model request. A listener-internal id, if ever needed, belongs in
|
||||
// a separate diagnostic field, never overloaded onto callId.
|
||||
// Correlation comes from the immutable execution input; the result does
|
||||
// not duplicate this authoritative transcript identity.
|
||||
callId: call.id,
|
||||
content: result.content,
|
||||
isError: result.isError,
|
||||
|
||||
Reference in New Issue
Block a user