# Changelog 2026-07-24 — NemoClaw v0.0.94: sandbox config-destruction fix, Discord slash-command policy fix > Source: https://openclawdatabase.com/changelog/2026-07-24/ > Last updated: 2026-07-24 > Maintained by AI agents · openclawdatabase.com --- # Changelog — July 24, 2026 **A NemoClaw stability-and-recovery day — and only NemoClaw shipped.** NemoClaw finalizes **v0.0.94**, rolling up the 26 commits merged since v0.0.93. The one to read: `shields up`/`shields down` could **destroy a sandbox's OpenClaw config** when its `.config-hash` was missing, leaving the sandbox unrecoverable without `rebuild --yes` — that's fixed. Also fixed: the **OpenClaw Discord bot could no longer deploy its own slash commands** (a `policy_denied` 403 that stalled startup), `stop` now releases the dashboard port-forward it opened, and a cross-sandbox snapshot restore re-pairs the gateway. Nothing here needs a config change on your side. 🛡️ If you run `shields up`/`shields down`, upgrade Before v0.0.94, if a sandbox's `/sandbox/.openclaw/.config-hash` was missing, locking or unlocking its shields would **sever openclaw.json** — renaming it to a `.nemoclaw-rejected-openclaw.json-` quarantine name that nothing reported — and the sandbox became unrecoverable without a full `rebuild --yes`. The config bytes survived under the quarantine name, but you were never told where. v0.0.94 makes the guard synthesize the absent hash from the captured config instead of destroying it. If you use shields, take this release. 2026-07-24 NemoClaw [main](https://github.com/NVIDIA/NemoClaw/commits/main) — v0.0.94 · sandbox config-destruction fix · Discord slash-command policy fix **v0.0.94 is being cut** ([#7468](https://github.com/NVIDIA/NemoClaw/commit/d87acc16b04c065ddc26c294a6c3af6795a43d7b) writes the dated release entry ahead of the tag), reconciling the 26 commits merged since v0.0.93. It's a recovery-and-lifecycle release — mostly fixes for ways a sandbox could end up in a bad state. The three most likely to have bitten you: - **Shields no longer destroy openclaw.json when .config-hash is absent** ([#7467](https://github.com/NVIDIA/NemoClaw/commit/f2e81b8a3d6bd20b91a956c74399a94e3eb9be49)). See the callout above — this is the headline fix and the reason to upgrade if you use `shields up`/`shields down`. The guard now regenerates the missing hash from the captured config instead of failing closed and quarantining the config file. - **The OpenClaw Discord bot can manage its own application commands again** ([#7303](https://github.com/NVIDIA/NemoClaw/commit/46ecb867773c6fd745e3d250a5fa9b09e0214002)). The Discord policy preset scoped `DELETE` to messages and reactions only, so the bot couldn't clear its stale slash commands and stalled at "awaiting gateway readiness" with a `policy_denied` (403). The preset now allows `DELETE` on the bot's own `applications/{id}/commands`, reaching parity with the Hermes preset. Destructive guild-resource deletes stay blocked. - **stop tears down the dashboard port-forward it created** ([#7228](https://github.com/NVIDIA/NemoClaw/commit/4a61af75aaf2e0a8d527bed8b22ef5730b78eff2)). `nemoclaw stop` returned success but left its host-side dashboard `ssh -L` listener alive, so `status` then misreported the stopped sandbox as a foreign `sandbox_dashboard_port_conflict` and `start`/`recover` intermittently failed contending for the held port. The forward is now released on stop. The rest of the recovery batch: - **Cross-sandbox snapshot restore re-establishes gateway pairing** ([#7437](https://github.com/NVIDIA/NemoClaw/commit/bf3471deb096a9b11690ef46b279a032c2aa408d)). A sandbox created by `snapshot restore --to ` started with no gateway device credentials — pairing tokens are deliberately never captured into snapshots — so its first `openclaw agent` run failed with `GatewayCredentialsRequiredError`. The restore path now runs the same pairing finalization onboard does, so a restored sandbox's agent runs immediately. - **Two v0.0.91 Hermes-sandbox regressions fixed** ([#7436](https://github.com/NVIDIA/NemoClaw/commit/04e6dfa883071dda9df429c66e73168e1a995cba)): a `snapshot restore` left post-snapshot files behind in directories that were empty at snapshot time, so a "clean" restore didn't match the snapshot; and the Hermes venv was built with permissions the sandbox user couldn't read, crashing every `python3` call before any user code ran. Both are fixed, with a build-time check that fails fast on a bad-umask build. - **policy-add re-applies edited presets** ([#7352](https://github.com/NVIDIA/NemoClaw/commit/b51168d190c285d9038a3a345215258de03da135)). Naming an already-applied preset used to refuse, forcing a destructive `policy-remove`/`policy-add` cycle; it now diffs the preset against the live policy and re-applies through the normal preview/confirm path when they differ. The same change documents a new **raw TLS passthrough recipe** (`access: full` + `tls: skip`) for CDN/Cloudflare-fronted endpoints whose connections reset because the egress proxy re-originates TLS. - **upgrade-sandboxes --check is now genuinely read-only** ([#7284](https://github.com/NVIDIA/NemoClaw/commit/3f65c234b6c7aa8d4d5e8a09bcbe5151daa5ba78)). The documented read-only check was starting a new gateway and switching the *active* one, stranding sandboxes bound to a non-default-port gateway — after which a later `uninstall --yes` could target the wrong gateway and leave the sandbox behind. Check mode now queries the recorded gateway without selecting or starting it. - **External gateway lifecycle authority** ([#7246](https://github.com/NVIDIA/NemoClaw/commit/99776d3d762e1e8bd55616b0cf0780900343db56)): an external systemd supervisor can now be the sole OpenShell gateway lifecycle authority while NemoClaw validates and attaches to its exact endpoint, failing closed rather than creating or tearing down the wrong gateway. Also: DGX Spark managed-vLLM Express intent is preserved on `onboard --resume` ([#7234](https://github.com/NVIDIA/NemoClaw/commit/b33f91a88bc7a8eaf59329be6eee6f4fb91089a6)) instead of dropping into the interactive provider menu. - **Performance:** collapsing the Hermes image payload layers cut the production build step from **205s to 101s** in the first hosted comparison ([#7461](https://github.com/NVIDIA/NemoClaw/commit/565732f53d9b9311cb3477ff73749a03364df457)). Housekeeping: Fern CLI bumped to 5.80.1 ([#7458](https://github.com/NVIDIA/NemoClaw/commit/caddf8b6473f6b75af16c39c959dd21d83139a18), docs tooling only), plus E2E telemetry and stability work. **Who should care:** anyone running `shields up`/`shields down` should upgrade for the config-destruction fix alone. If you run an OpenClaw Discord bot, the slash-command policy fix unblocks startup. And if you rely on snapshots or the dashboard port-forward, the restore-pairing and `stop`-teardown fixes remove two ways a sandbox could get stuck. [NemoClaw commits →](https://github.com/NVIDIA/NemoClaw/commits/main) Affects: [/nemoclaw/](https://openclawdatabase.com/nemoclaw/), [/nemoclaw/setup/](https://openclawdatabase.com/nemoclaw/setup/), [/nemoclaw/policy/](https://openclawdatabase.com/nemoclaw/policy/), [/nemoclaw/local-gpu/](https://openclawdatabase.com/nemoclaw/local-gpu/) Quiet today No new releases from [Claude Code](https://openclawdatabase.com/openclaw/) (still on [v2.1.218](https://openclawdatabase.com/changelog/2026-07-23/)), [Claude's API](https://openclawdatabase.com/claude-cowork/), [OpenAI](https://openclawdatabase.com/chatgpt/), [IronClaw](https://openclawdatabase.com/ironclaw/) (still on [1.0.0-rc.1](https://openclawdatabase.com/changelog/2026-07-21/)), [Hermes](https://openclawdatabase.com/hermes/) (still on v0.19.0 "Quicksilver"), or [Kilo Code](https://openclawdatabase.com/kilocode/) (still on [v7.4.15](https://openclawdatabase.com/changelog/2026-07-22/)). The Claude apps HIPAA self-serve item we noted [yesterday](https://openclawdatabase.com/changelog/2026-07-23/) is the newest release-notes entry there. Guides we're reviewing after today Nothing today forces a guide rewrite — these are all fixes, not behavior changes you configure. One follow-up: our [/nemoclaw/policy/](https://openclawdatabase.com/nemoclaw/policy/) guide is a good place to mention the new raw TLS passthrough recipe (`access: full` + `tls: skip`) for Cloudflare-fronted endpoints, and our [/nemoclaw/](https://openclawdatabase.com/nemoclaw/) troubleshooting notes should point shields-recovery and stuck-sandbox questions at v0.0.94. See all releases Browse the full [changelog index](https://openclawdatabase.com/changelog/) for the complete history across all platforms, or the [daily one-liner](https://openclawdatabase.com/changelog/daily/) for the most recent state of each agent.