@mindstudio-ai/agent 0.1.101 → 0.1.102
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +6 -0
- package/dist/cli.js +30 -16
- package/dist/index.d.ts +44 -0
- package/dist/index.js +41 -11
- package/dist/index.js.map +1 -1
- package/dist/postinstall.js +30 -16
- package/llms.txt +3 -3
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -436,6 +436,12 @@ input a careful operator would submit (or abstains with `input: null`), gets gra
|
|
|
436
436
|
against what a human actually did, and climbs an autonomy ladder (`shadow` → `approve` →
|
|
437
437
|
`auto`) declared in the app manifest.
|
|
438
438
|
|
|
439
|
+
Attach the deciding task's transcript to the proposal: `return { input, reasoning,
|
|
440
|
+
trace: task.traceId }` (an array for escalation chains). The platform records
|
|
441
|
+
jewel-run transcripts server-side per turn; the attached id preserves the full model
|
|
442
|
+
transcript with the pair — the training row and the audit trail. A custom `grade`'s
|
|
443
|
+
verdict may attach its judge's `trace` the same way.
|
|
444
|
+
|
|
439
445
|
For decision moments the app detects itself (an ingest branch that lands a row in a
|
|
440
446
|
pending state), hand the moment to the jewels from backend code:
|
|
441
447
|
|