Skip to content

v0.0.140

@rsliter rsliter tagged this 29 Oct 23:49
## Release range

- Previous release: `v0.0.139` at `a08ae7aa911ef5bd7a4b72d9020f575d75d0adb1`
- Candidate: `88fe14f3fb9a0bf72997cce19ec88b5b2ee08391`
- Candidate selection: current-main
- Commits: 61
- Risky files detected: 152

## QA context

### Risky areas

- Workflow / enforcement
- Sandbox / policy / SSRF
- Installer / bootstrap
- Credentials / inference
- Onboarding / host glue

### Suggested test focus

- Fresh install and upgrade paths
- Onboarding wizard and sandbox creation
- Policy enforcement, network egress, and SSRF protections
- CI checks, pre-commit hooks, and DCO declarations
- Credential storage and inference provider routing

## Canonical release entry

- Path: `docs/changelog/2026-10-29.mdx` at the planned candidate
- Entry:

## v0.0.140

NemoClaw v0.0.140 improves sandbox recovery, onboarding and inference checks, Hermes credentials, and managed Deep Agents Code. It also makes backup, uninstall, and installer failures easier to recover from.

- The managed Deep Agents Code image now uses version 0.1.87. Its initial model request retains `search_tools` when an upstream request projection omits that tool, so managed MCP search remains available. Related changes: [PR #13878](https://github.com/NVIDIA/NemoClaw/pull/13878) and [PR #13766](https://github.com/NVIDIA/NemoClaw/pull/13766).
- Hermes images now include Switchyard 0.3.0 for qualification. The package remains disabled by default and does not enable a new route. Related change: [PR #13879](https://github.com/NVIDIA/NemoClaw/pull/13879).
- Hermes rebuilds preserve managed OpenRouter credential references while removing raw or unrecognized references from restored native state. Hermes also accepts the managed NVIDIA key template when its native credential slots agree. Related changes: [PR #13771](https://github.com/NVIDIA/NemoClaw/pull/13771) and [PR #13785](https://github.com/NVIDIA/NemoClaw/pull/13785).
- Hermes restores the default profile's `SOUL.md` and skills index. When neither Tavily nor the Nous-managed web gateway is configured, Hermes does not offer `web_search` or `web_extract`. Related changes: [PR #12386](https://github.com/NVIDIA/NemoClaw/pull/12386) and [PR #13620](https://github.com/NVIDIA/NemoClaw/pull/13620).
- Hermes configuration export preserves a disabled dashboard and omits its port, which v1 rejects when the dashboard is disabled. It validates retained dashboard URLs against the allocated port. Related change: [PR #12813](https://github.com/NVIDIA/NemoClaw/pull/12813).
- A model-only change on a recorded DNS-backed HTTPS adapter route now verifies the selected model from the sandbox before it updates agent configuration or registry state. If verification fails, NemoClaw attempts to restore the previous OpenShell model selection and reports the result. Related change: [PR #13237](https://github.com/NVIDIA/NemoClaw/pull/13237).
- Local inference onboarding uses `NEMOCLAW_LOCAL_INFERENCE_TIMEOUT` to bound each applicable in-sandbox request, with a 60-second minimum. Managed llama.cpp fingerprint requests have a longer bounded deadline than operator-managed servers. Related changes: [PR #13670](https://github.com/NVIDIA/NemoClaw/pull/13670) and [PR #13561](https://github.com/NVIDIA/NemoClaw/pull/13561).
- The Ollama model menu preserves a reported `0` free-GPU-memory reading instead of treating it as unknown. Managed vLLM publishes provider keys atomically, so a failed update does not leave a partial key set. Related changes: [PR #13748](https://github.com/NVIDIA/NemoClaw/pull/13748) and [PR #13137](https://github.com/NVIDIA/NemoClaw/pull/13137).
- NVIDIA endpoint validation during onboarding or resume uses the native endpoint when saved metadata still names `inference.local` or has no endpoint. A provider with only a binding and no ownership receipt is treated as detached. Related changes: [PR #13571](https://github.com/NVIDIA/NemoClaw/pull/13571) and [PR #13649](https://github.com/NVIDIA/NemoClaw/pull/13649).
- If onboarding finds a native provider without a NemoClaw ownership receipt, its error gives an `openshell provider delete` command with the gateway flag when needed. Check that no sandbox on that gateway uses the provider before running it. Related change: [PR #13669](https://github.com/NVIDIA/NemoClaw/pull/13669).
- Onboarding preserves custom endpoint URL, API family, and credential variable name at the configuration review checkpoint. It rejects unsupported Deep Agents Code auto-approval choices before startup and rejects invalid `NEMOCLAW_CPU` or `NEMOCLAW_RAM` percentages. Related changes: [PR #13779](https://github.com/NVIDIA/NemoClaw/pull/13779), [PR #13567](https://github.com/NVIDIA/NemoClaw/pull/13567), and [PR #12422](https://github.com/NVIDIA/NemoClaw/pull/12422).
- Sandbox `status`, `recover`, and `connect --probe-only` identify the target gateway. Status stays responsive during rebuild and marks inference as not probed when a safe read-only lifecycle snapshot is available. Related changes: [PR #13134](https://github.com/NVIDIA/NemoClaw/pull/13134) and [PR #13009](https://github.com/NVIDIA/NemoClaw/pull/13009).
- `destroy` sends cleanup to the sandbox's recorded gateway without changing the active gateway selection. Sandboxes on other gateways do not block final cleanup of the target gateway. Related changes: [PR #13012](https://github.com/NVIDIA/NemoClaw/pull/13012) and [PR #13819](https://github.com/NVIDIA/NemoClaw/pull/13819).
- `backup-all` retries native-home capture when a stopped sandbox is still booting and the capture reports process churn. Other capture failures are not retried. Snapshot sanitization also accepts backslashes in recorded paths. Related changes: [PR #12891](https://github.com/NVIDIA/NemoClaw/pull/12891) and [PR #12381](https://github.com/NVIDIA/NemoClaw/pull/12381).
- Deep Agents Code image builds now keep their npm cache outside `/sandbox`, so a restored agent home's `.npm` state does not collide with the build cache. Related change: [PR #13263](https://github.com/NVIDIA/NemoClaw/pull/13263).
- Uninstall removes a dead gateway service on headless Linux and skips an absent default gateway during a whole-host sweep. It reports when the container engine does not respond. Related changes: [PR #13630](https://github.com/NVIDIA/NemoClaw/pull/13630), [PR #13446](https://github.com/NVIDIA/NemoClaw/pull/13446), and [PR #13645](https://github.com/NVIDIA/NemoClaw/pull/13645).
- The installer distinguishes a timed-out release lookup, an unavailable requested reference, and a Git verification failure.
  A managed install or upgrade can remove the existing CLI source checkout before a failed lookup; fix the lookup issue and rerun the installer to restore the command.
  Related change: [PR #11531](https://github.com/NVIDIA/NemoClaw/pull/11531).
- Deep Agents Code edit approval cards show the full requested path for a nonempty `edit_file` path. NemoClaw writes `POLICY.md` only in OpenClaw sandboxes. Related changes: [PR #13591](https://github.com/NVIDIA/NemoClaw/pull/13591) and [PR #13713](https://github.com/NVIDIA/NemoClaw/pull/13713).
- A piped `nemoclaw <name> agent` message must finish within its host-side input deadline and be no more than 4 MiB. A timeout, oversized message, or read failure exits without starting an agent turn. Related change: [PR #13646](https://github.com/NVIDIA/NemoClaw/pull/13646).
- Ordinary installer and supported CLI operations attempt to send telemetry about completed operations to NVIDIA's fixed TEST receiver by default. Production delivery is not enabled. Set `NEMOCLAW_DISABLE_TELEMETRY=1` in each shell before installation or CLI use to disable it. Related change: [PR #13772](https://github.com/NVIDIA/NemoClaw/pull/13772).
- CI operations now attempt to send telemetry to the TEST receiver with a source-run label.
  If GitHub Actions environment values pass format checks, the label includes their reported run ID and attempt.
  Other eligible CI runs use a generic label.
  Opt-out, malformed explicit QA labels, and test-process suppression still apply.
  Related change: [PR #13770](https://github.com/NVIDIA/NemoClaw/pull/13770).

## Documentation coverage

- Latest included cumulative docs PR: [#13673](https://github.com/NVIDIA/NemoClaw/pull/13673), merged 2026-10-26, final PR head `dc3e19bff134bcca605e879e32291aff69024fcd`.
- Final PR commit and merge commit: `dc3e19bff134bcca605e879e32291aff69024fcd` and `a08ae7aa911ef5bd7a4b72d9020f575d75d0adb1`.
- Final automated refresh coverage commit: `14c9f5982b7678e6279c32899120657ab5a92bd6`, the last verified bot refresh commit's main parent and an ancestor of this candidate.
- Later commits and merged PRs: 68 first-parent commits after the automated refresh point; each maps to a merged PR:
  - `33819def84f7ba80256f7ae6dc014a775aaa6ab9` [#13717](https://github.com/NVIDIA/NemoClaw/pull/13717) fix(hermes): pin the base with Node 24.22.1
  - `469591818feffee15bedf8704359298a6e47a399` [#12845](https://github.com/NVIDIA/NemoClaw/pull/12845) fix(docs): report post-merge draft conflicts
  - `27aa219854271a2bea893f64d4af17c3e1eeb0bc` [#13642](https://github.com/NVIDIA/NemoClaw/pull/13642) fix(inference): classify transient NVIDIA probe failures
  - `bc5e649e3e7f4bec6b3d251014cf3c66dd2135f2` [#13680](https://github.com/NVIDIA/NemoClaw/pull/13680) fix(rebuild): resume after absent upgrade replacement
  - `dfe00e9ccaa814632d39baed21a20f7b9f495f6d` [#13650](https://github.com/NVIDIA/NemoClaw/pull/13650) fix(onboard): show native local policy fix and rerun inference on resume
  - `143678b745b0f558c957559dee26f997588b2c5e` [#13596](https://github.com/NVIDIA/NemoClaw/pull/13596) fix(uninstall): remove managed OpenClaw state volumes
  - `a08ae7aa911ef5bd7a4b72d9020f575d75d0adb1` [#13673](https://github.com/NVIDIA/NemoClaw/pull/13673) docs: prepare v0.0.139 documentation
  - `868628351585b8e251d04ab81a60ad832d6550e0` [#13635](https://github.com/NVIDIA/NemoClaw/pull/13635) fix(e2e): qualify native OpenShell managed runtime
  - `9347d8928be41deaa78febfe34dc357839251e6b` [#13748](https://github.com/NVIDIA/NemoClaw/pull/13748) fix(inference): keep a zero free-memory reading for Ollama
  - `04d2e83db973a1c0efebd40a7ea490a42cea9403` [#13648](https://github.com/NVIDIA/NemoClaw/pull/13648) docs(status): state liveRoute is null for native provider sandboxes
  - `edefa91ae40aa4bff8140766352850578a059018` [#13696](https://github.com/NVIDIA/NemoClaw/pull/13696) fix(hermes): back off archive rate limits
  - `9d35b77d2155c17b966f78aa2c359a9807a41d7c` [#13630](https://github.com/NVIDIA/NemoClaw/pull/13630) fix(uninstall): remove dead gateway unit on headless Linux
  - `159fbf46014e06dda39fdcf104cbbe498345eb23` [#13772](https://github.com/NVIDIA/NemoClaw/pull/13772) docs: use approved operation telemetry notice
  - `528f4c4ce5a18f944adf309c2d7ed34763a7d32f` [#13620](https://github.com/NVIDIA/NemoClaw/pull/13620) fix(hermes): disable keyless web search fallback without a provider
  - `e5a79fe5e1ad5a329187ebf242bef273ad021135` [#12813](https://github.com/NVIDIA/NemoClaw/pull/12813) fix(config): export disabled Hermes dashboards on allocated ports
  - `39af83a926759a9fcb701b9dfa6ad3bafe7af09e` [#13446](https://github.com/NVIDIA/NemoClaw/pull/13446) fix(uninstall): skip absent default gateway in whole-host sweep
  - `a6e6d233e26926ed6e6b5fda9a8b468482c05dc6` [#13137](https://github.com/NVIDIA/NemoClaw/pull/13137) fix(inference): publish managed vLLM keys atomically
  - `143fe465abedf33ea49ed769c284213f5457bb90` [#13646](https://github.com/NVIDIA/NemoClaw/pull/13646) fix(agent): enforce the piped-input deadline on the host
  - `5aa013f9808b9ec831aa979b06d89415d3d47622` [#13753](https://github.com/NVIDIA/NemoClaw/pull/13753) fix(ci): remove open PR caps for four contributors
  - `edf1cd3318b7c096953036f050ad61633bc000ca` [#13591](https://github.com/NVIDIA/NemoClaw/pull/13591) fix(dcode): preserve full edit approval paths
  - `f5afe269c08fb88b53f1c17c28daa74985015937` [#13567](https://github.com/NVIDIA/NemoClaw/pull/13567) fix(onboard): reject unsupported auto-approval before startup
  - `4aa520a5048a18c95c9eb6fd7fbcdfaafc621dff` [#13134](https://github.com/NVIDIA/NemoClaw/pull/13134) fix(sandbox): report gateway targets in status and recovery
  - `8712ceaaa7854e8d1bf76abf41fc71d882f3c1f8` [#13237](https://github.com/NVIDIA/NemoClaw/pull/13237) fix(inference): verify reused HTTPS adapter routes from the sandbox
  - `3ce2366d5e62e2b96d5a637c225f33677ccaa161` [#13763](https://github.com/NVIDIA/NemoClaw/pull/13763) fix(e2e): recognize reviewed OpenShell 0.1.5 runtime
  - `bafe6ccdee09551d5394f5e8bbc49a4830577b42` [#13645](https://github.com/NVIDIA/NemoClaw/pull/13645) fix(uninstall): say when the container engine is not responding
  - `7dbdf4e7af7e458b5786c4caf4d61efa1005dc5b` [#12386](https://github.com/NVIDIA/NemoClaw/pull/12386) fix(hermes): restore the default profile's SOUL.md and skills index
  - `5f81921a04375a043e6a20ee024f5dee66308689` [#11531](https://github.com/NVIDIA/NemoClaw/pull/11531) fix(installer): clarify release lookup failures
  - `1204c3d63e760e107a830f466e77bfbeeeabb3e3` [#13713](https://github.com/NVIDIA/NemoClaw/pull/13713) fix(policy): write POLICY.md only in OpenClaw sandboxes
  - `cb5dc001a7601a39a79ec470159f0abe6d1fe2aa` [#13670](https://github.com/NVIDIA/NemoClaw/pull/13670) fix(onboard): honor the local inference timeout in the sandbox inference check
  - `87cc4c8dd531038fed89961b4311b83f1436576b` [#12899](https://github.com/NVIDIA/NemoClaw/pull/12899) fix(onboard): suppress false fresh-onboard warnings
  - `1845e56ccae8e6655d793a555415dfa79c173944` [#13787](https://github.com/NVIDIA/NemoClaw/pull/13787) ci(e2e): provision supported Python for Model Router
  - `1d23da934a95a9b78fbc4a35c2529f1815debc6b` [#13796](https://github.com/NVIDIA/NemoClaw/pull/13796) fix(e2e): prepare reviewed SDK artifacts before compilation
  - `dff9a02eb3e7c7354ef047db148b342ca0eb5bc9` [#13009](https://github.com/NVIDIA/NemoClaw/pull/13009) fix(status): keep status responsive during rebuild
  - `1173431d76c8a0fb5e0e2cf9e434fdd23e8e331e` [#13766](https://github.com/NVIDIA/NemoClaw/pull/13766) fix(deepagents): keep managed MCP search available
  - `357da69e65c391b90f438a3e9ed082cd00012c02` [#13762](https://github.com/NVIDIA/NemoClaw/pull/13762) fix(e2e): retry timed-out Pi read turn
  - `cf421b6db433b862559b65aa2b341c3db577c468` [#13682](https://github.com/NVIDIA/NemoClaw/pull/13682) fix(telemetry): preserve bounded observed schema values
  - `34ab747a1f39a7ede9e62f33ae249eb39ec7ff21` [#12422](https://github.com/NVIDIA/NemoClaw/pull/12422) fix(onboard): reject invalid NEMOCLAW_CPU and NEMOCLAW_RAM percentages
  - `8c72b552cc92fe7082908f3a13db7dc94f2f6685` [#13779](https://github.com/NVIDIA/NemoClaw/pull/13779) fix(onboard): keep custom route metadata in the review checkpoint
  - `f47efaebdaa8c981d660e261cd1e79b22e92a5f1` [#13785](https://github.com/NVIDIA/NemoClaw/pull/13785) fix(hermes): accept the NVIDIA key template as a current credential
  - `d7137496475dd9353053001cf58f09eddfdd7e05` [#13012](https://github.com/NVIDIA/NemoClaw/pull/13012) fix(sandbox): preserve active gateway on destroy
  - `90603b99fe21a770fc586e4b38869e6e1d73a051` [#13669](https://github.com/NVIDIA/NemoClaw/pull/13669) fix(inference): name the command that clears an unowned native provider
  - `2d085490b914490d921634fdc58ce9fff40a8026` [#13649](https://github.com/NVIDIA/NemoClaw/pull/13649) fix(providers): treat binding-only providers as detached
  - `4c96c1087f49fb7d5b1b94dc4f90081ec34a2298` [#13771](https://github.com/NVIDIA/NemoClaw/pull/13771) fix(hermes): preserve OpenRouter authentication across rebuild
  - `1c15109612b2983cd54cbee1a202cbf9ddd2f3ce` [#13571](https://github.com/NVIDIA/NemoClaw/pull/13571) fix(onboard): use native NVIDIA endpoint for resume smoke
  - `eeb148322f7f6092eb6fecd3c8c1b3c6eb49a547` [#13561](https://github.com/NVIDIA/NemoClaw/pull/13561) fix(llama-cpp): allow managed bridge fingerprint latency
  - `1864afdba73f5cbbed86972a7660183d2f59a837` [#13697](https://github.com/NVIDIA/NemoClaw/pull/13697) ci(e2e): prepare trusted OpenClaw 10.5 qualification
  - `06f976dc68d4c29595ec5a962fbcc6a0e75a6e1d` [#12891](https://github.com/NVIDIA/NemoClaw/pull/12891) fix(backup): retry stopped-sandbox capture while the boot settles (Fixes #12867)
  - `e234ef62a5cae2c8f709ae9dbd4a0cd8c710046d` [#13023](https://github.com/NVIDIA/NemoClaw/pull/13023) fix(dev-setup): prevent heap and prerequisite setup failures
  - `3619e535e2a63bd6e046e990a04710dda8db2443` [#13263](https://github.com/NVIDIA/NemoClaw/pull/13263) fix(sandbox): isolate npm upgrade cache from native state
  - `bdcd7869ddf0ed84b07e6d1ec42b24a8200406d4` [#12381](https://github.com/NVIDIA/NemoClaw/pull/12381) fix(sandbox): accept backslashes in snapshot sanitizer paths
  - `32c42bbad69dca3bb3f4e11ea439c8ce4898d4ff` [#13857](https://github.com/NVIDIA/NemoClaw/pull/13857) perf(e2e): reuse equivalent OpenClaw base in protected runs
  - `7ced2348dbefa938581efb8fe576f3e83cfa653c` [#13876](https://github.com/NVIDIA/NemoClaw/pull/13876) feat(contributing): report authoring sessions in pull requests
  - `be1a9683a02e73ab5f66b061c0955e2c59319e0b` [#13875](https://github.com/NVIDIA/NemoClaw/pull/13875) fix(ci): compare growth against the proposed merge (Refs #13809)
  - `f568a3f4935d6a467f137690ffe6c5868b1ae632` [#13887](https://github.com/NVIDIA/NemoClaw/pull/13887) fix(e2e): keep inference receipt authority on the host
  - `816dafef5a9c3669c87c3b5c34aefee1429245a2` [#13819](https://github.com/NVIDIA/NemoClaw/pull/13819) fix(services): stop calling a valid sandbox name invalid on destroy
  - `37f1e4cb9942a3dc05c0c76995f86624261bc7b1` [#13930](https://github.com/NVIDIA/NemoClaw/pull/13930) fix(ci): avoid deep growth history fetches for recent PRs
  - `ef3ffec452ba573618df05675635f8e1605b60ed` [#13932](https://github.com/NVIDIA/NemoClaw/pull/13932) fix(e2e): retain bounded OpenClaw answer diagnostics
  - `670afc947418241eea7ef2bc1ccf7988698be351` [#13879](https://github.com/NVIDIA/NemoClaw/pull/13879) feat(hermes): package released Switchyard 0.3.0
  - `1bba415e52ffd1efff22095d95ba1962cee12ffb` [#13908](https://github.com/NVIDIA/NemoClaw/pull/13908) docs(inference): correct vLLM registry login and variant evidence (#13894)
  - `5d463e0bc1c2d625abc823c10137fb738ecf198a` [#13878](https://github.com/NVIDIA/NemoClaw/pull/13878) chore(deps): update LangChain Deep Agents Code to 0.1.87
  - `b10117ecdf3318da48e882362d971dce6879750c` [#13629](https://github.com/NVIDIA/NemoClaw/pull/13629) fix(rebuild): scope the Hermes API port to each rebuild
  - `015f5f703d2b6940c90cea2a42cdeafab47803f8` [#13984](https://github.com/NVIDIA/NemoClaw/pull/13984) test(e2e): set isolated HOME for native credential rejection
  - `18378d10eae0cc3ea5dd503fd9c73595b073e209` [#13686](https://github.com/NVIDIA/NemoClaw/pull/13686) ci(e2e): retain portable Hermes failure diagnostics and cleanup
  - `51b14d3650f9022e9f95e25cd5559b916256660b` [#13993](https://github.com/NVIDIA/NemoClaw/pull/13993) docs: prepare v0.0.140 documentation
  - `e677f436a8f1ada047c0fa76bdd37bcac58af244` [#13998](https://github.com/NVIDIA/NemoClaw/pull/13998) docs: correct v0.0.140 installer lookup guidance
  - `287ebd82eb52b9b796ac0bbb134ab06b5fa72bc2` [#13770](https://github.com/NVIDIA/NemoClaw/pull/13770) fix(telemetry): identify CI operation records
  - `f12cc4806a780b6a29748896356cabd857cf6631` [#13742](https://github.com/NVIDIA/NemoClaw/pull/13742) fix(onboard): reuse a Ready sandbox on resume instead of recreating it for reasoning drift (#13732)
  - `88fe14f3fb9a0bf72997cce19ec88b5b2ee08391` [#14001](https://github.com/NVIDIA/NemoClaw/pull/14001) docs(changelog): note CI telemetry in v0.0.140
- Changed paths: #13673 touched only `docs/changelog/2026-10-26.mdx`, `docs/manage-sandboxes/recover-rebuild-sandboxes.mdx`, `docs/reference/system-readiness.mdx`, and `docs/reference/troubleshooting.mdx`; all are allowed documentation paths. [#13993](https://github.com/NVIDIA/NemoClaw/pull/13993), [#13998](https://github.com/NVIDIA/NemoClaw/pull/13998), and [#14001](https://github.com/NVIDIA/NemoClaw/pull/14001) prepared the release entry. [#13770](https://github.com/NVIDIA/NemoClaw/pull/13770) updated the owning CI telemetry guidance. [#13742](https://github.com/NVIDIA/NemoClaw/pull/13742) changed only onboarding code and tests; the owning capability page already describes reuse of recorded reasoning values on resume, while the release entry does not name that fix.
- Review and checks: #13673 APPROVED; 27 successful checks, 25 legitimately skipped, and one cancelled duplicate `commit-lint` with a successful run also present. #13993, #13998, #13770, #13742, and #14001 were approved and merged. #14001 had 42 successful and 28 skipped checks, plus one cancelled duplicate and two context entries without a conclusion in the final rollup; required branch checks passed. Its Advisor specialists completed clear with no additional E2E recommendations, and its automatic exact-commit E2E passed.
- Open managed docs PRs: [#13774](https://github.com/NVIDIA/NemoClaw/pull/13774), head `fd17f1b24755549a65abb9915ee9dbbf881d2536`, superseded by #13993 and awaiting a separate close decision.
- Maintainer decision: Proceed with the candidate as shown.

## Base and managed image evidence

- Base-image candidate: `88fe14f3fb9a0bf72997cce19ec88b5b2ee08391`
- Candidate [E2E attempt 2](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/attempts/2) failed its [base-image-publication job](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/job/123744734806) because the applicable [f12cc480 publisher run](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574) was cancelled after a newer main push. The failure is publication availability, not a candidate test assertion.
- Applicable [f12cc480 publisher attempt 2](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/attempts/2) failed only its [Hermes amd64 qualification](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/job/123763488653): the clean root-entrypoint Switchyard package contract exited 139 (segmentation fault) during native activation. Five other retained root-entrypoint scenarios passed. The newer [de009775 publisher](https://github.com/NVIDIA/NemoClaw/actions/runs/40954765481) passed its Hermes amd64 qualification, and the intervening changed paths contain no Hermes image or test files; this comparison suggests an intermittent native failure but does not qualify the planned candidate.
- [Publisher attempt 3](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/attempts/3) passed all six Hermes root-entrypoint smoke tests, then failed the publication evidence step: the failed-job rerun retained cohort `ghrun-40951600574-2` while the job ran in attempt 3. This did not reproduce the native segmentation fault and did not complete publication.
- Rebecca approved a full rerun. [Publisher attempt 4](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/attempts/4) completed successfully on `f12cc4806a780b6a29748896356cabd857cf6631`, including [Hermes amd64](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/job/123787780226), [Hermes arm64](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/job/123787780317), and [multi-platform managed-image promotion](https://github.com/NVIDIA/NemoClaw/actions/runs/40951600574/job/123792430524). At that point, the planned candidate still required a successful exact-SHA `base-image-publication` verifier job.
- Rebecca declined a full candidate E2E rerun and chose to rely on the earlier full E2E run for general E2E context. She separately approved a targeted rerun of only the independent `base-image-publication` job after the publisher passed.
- Evidence: Candidate [E2E workflow attempt 3](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/attempts/3), `push` on `main` at exact candidate `88fe14f3fb9a0bf72997cce19ec88b5b2ee08391`, completed successfully. It produced a successful [`base-image-publication` job](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/job/123798343429), job ID `123798343429`, completed 2026-10-29T23:37:49Z. The successful candidate check run, job metadata, and owning workflow attempt were verified. Downstream [Jetson](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/job/123798865537) and [scorecard](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/job/123801475336) jobs also passed.

## General E2E decision

- Newest identifiable full manual main run at inspection 2026-10-29T23:41:51Z: [40912376834](https://github.com/NVIDIA/NemoClaw/actions/runs/40912376834), attempt 1, SHA `5d463e0bc1c2d625abc823c10137fb738ecf198a` (older than planned candidate `88fe14f3fb9a0bf72997cce19ec88b5b2ee08391`); created and started 2026-10-29T16:31:12Z, last updated 2026-10-29T18:03:54Z, age 7 hours 10 minutes, completed successfully. [Release qualification](https://github.com/NVIDIA/NemoClaw/actions/runs/40912376834/job/123634441993) passed from 18:03:14Z to 18:03:19Z, and [Exact staging Brev Launchable](https://github.com/NVIDIA/NemoClaw/actions/runs/40912376834/job/123587842298) passed from 17:32:33Z to 18:01:14Z. Sixteen non-applicable jobs were skipped; none failed, was cancelled, or remained active. No newer run titled `E2E full main` appeared in the latest 100 manual main runs.
- Eight first-parent main commits landed between the full-run SHA and this candidate: three behavior fixes (Hermes rebuild, CI telemetry, Ready-sandbox reuse), two E2E test/CI changes, and three release-documentation changes. Current main `de00977585d04f3bac7b090a0870b0d2dca4972d` adds two more behavior fixes after the frozen candidate, for ten first-parent commits after the full-run SHA.
- The targeted candidate verifier rerun also started downstream [jetson-nvmap-gpu](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/job/123798865537), which passed at 2026-10-29T23:45:00Z, and [scorecard](https://github.com/NVIDIA/NemoClaw/actions/runs/40951677326/job/123801475336), which passed. The attempt completed successfully with no failed or active jobs; it was a targeted rerun, not a new full E2E suite.
- Requested focused or full release runs: None yet.
- Maintainer E2E choice: Proceed with the status as shown.

Exceptions: I accept the eight-commit gap, including three behavior changes and two E2E test/CI changes, because the earlier full E2E passed today and the candidate image verifier, Jetson job, and scorecard passed.
Assets 2
Loading