You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: .agents/skills/nemoclaw-maintainer-e2e/SKILL.md
+72-20Lines changed: 72 additions & 20 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -11,7 +11,8 @@ description: Dispatches and verifies trusted GitHub Actions E2E for NemoClaw mai
11
11
Use `.github/workflows/e2e.yaml` from trusted `main`.
12
12
Every push to `main` selects the default workflow E2E jobs.
13
13
Push runs skip `jetson-nvmap-gpu`, `llama-cpp-dgx-spark-plan`, and `llama-cpp-dgx-spark-qualification` because push events cannot set the required workflow dispatch flags.
14
-
Pre-tag evidence still requires the full `workflow_dispatch` mode described below.
Reuse `run-$RUN_ID.json` and `jobs-$RUN_ID.json` as the `nemoclaw-maintainer-cut-release-tag` manifest inputs and as the full-mode validator inputs. Do not fetch the same run again. `jobs-latest-$RUN_ID.json` is only for ordinary and Launchable modes.
322
+
Reuse `run-$RUN_ID.json` and `jobs-$RUN_ID.json` as the
323
+
`nemoclaw-maintainer-cut-release-tag` manifest inputs when building a release ledger and as validator
324
+
inputs when validating image evidence. Do not fetch the same run again. Use
325
+
`jobs-latest-$RUN_ID.json` only for an ordinary status report that does not validate image evidence.
317
326
318
327
For ordinary and Launchable modes, require `run-$RUN_ID.json` to report:
319
328
@@ -324,7 +333,8 @@ For ordinary and Launchable modes, require `run-$RUN_ID.json` to report:
324
333
For Launchable mode, also require `jobs-latest-$RUN_ID.json` to contain one completed, successful
325
334
`Exact staging Brev Launchable` job. Return the workflow and job URLs.
326
335
327
-
For full mode, select and download the Launchable E2E artifact for the latest successful Launchable job attempt:
336
+
For full mode, selective Launchable mode, or a trusted `main` push selected for an image handoff,
337
+
download the artifact for the latest successful Launchable job attempt:
Leave `EVIDENCE_MODE` unset for the legacy full-run validator path. For a release-image evidence
370
+
candidate, set it to `full`, `launchable`, or `push`. That stricter path rejects PR-shaped evidence.
371
+
372
+
Every validator mode requires:
354
373
355
374
- the workflow run to succeed for the selected SHA;
356
-
- `dispatch.json` to bind the same run, empty selectors, `include_staging_brev_launchable=true`, `allowJetsonRunnerQueue: false`, `allowDgxSparkRunnerQueue: false`, and the selected successful Launchable job attempt;
375
+
- `dispatch.json` to bind the same runand selected successful Launchable job attempt;
357
376
- `allowJetsonDispatch: false` in every v2 `dispatch.json` receipt;
358
377
- `allowJetsonDispatch` to be absent or `false` in every v1 `dispatch.json` receipt;
378
+
- `allowJetsonRunnerQueue: false` and `allowDgxSparkRunnerQueue: false`;
359
379
- `Exact staging Brev Launchable` to conclude `success` in the selected current or earlier attempt of the same workflow run;
360
380
- `launchable-e2e.json` to identify the selected SHA in the repository and provision records;
381
+
- `launchable-e2e.json` to bind the booted GCP image URI to its project, name, numeric ID,
382
+
self-link, source family, image origin, creation time, and image-repository SHA;
383
+
- the source project to equal trusted project `brevdevprod`;
361
384
- the booted repository to be unmodified;
362
385
- the in-guest full E2E to pass; and
363
386
- `cleanup.json` to report the same workspace as `ABSENT`.
364
387
388
+
The selector contract is mode-specific: `full` requires empty selectors and
0 commit comments