fix(scripts): allow CLI clients in auto-pair watcher by stevenrick · Pull Request #1496 · NVIDIA/NemoClaw

stevenrick · 2026-04-05T02:31:47Z

Summary

The auto-pair watcher in nemoclaw-start.sh rejects CLI device-pairing requests, making the OpenClaw gateway unreachable for all openclaw CLI commands inside the sandbox (channels, status, tui, devices, etc.). Adding 'cli' to ALLOWED_MODES fixes this.

Related Issue

Fixes #1310

Related: #1310

Changes

Add 'cli' to ALLOWED_MODES in the auto-pair watcher embedded in scripts/nemoclaw-start.sh

Type of Change

Code change for a new feature, bug fix, or refactor.
Code change with doc updates.
Doc only. Prose changes without code sample modifications.
Doc only. Includes code sample changes.

Testing

Tested on a Brev instance with a fresh nemoclaw onboard:

Before: openclaw channels status → gateway connect failed: GatewayClientRequestError: pairing required

After: openclaw channels status → Gateway reachable.

openclaw devices list --json confirms the CLI client is paired with full operator scopes. The existing security comment notes this allowlist is "defense-in-depth, not a trust boundary" — the sandbox user can already execute openclaw agent --local without gateway access.

npx prek run --all-files passes (or equivalently make check).
npm test passes.
make docs builds without warnings. (for doc-only changes)

Checklist

General

I have read and followed the contributing guide.
I have read and followed the style guide. (for doc-only changes)

Code Changes

Formatters applied — npx prek run --all-files auto-fixes formatting (or make format for targeted runs).
Tests added or updated for new or changed behavior.
No secrets, API keys, or credentials committed.
Doc pages updated for any user-facing behavior changes (new commands, changed defaults, new features, bug fixes that contradict existing docs).

Signed-off-by: Steven Rick srick@nvidia.com

Summary by CodeRabbit

New Features
- Expanded device support to include CLI mode in auto-pairing functionality.

Add 'cli' to ALLOWED_MODES so sandbox-internal CLI commands can connect to the gateway. Related: #1310 Signed-off-by: Steven Rick <srick@nvidia.com>

coderabbitai · 2026-04-05T02:32:01Z

📝 Walkthrough

Walkthrough

The auto-pair watcher's client-mode allowlist is expanded to accept both 'webchat' and 'cli' client modes, replacing the single-mode condition. All other control flow, security checks, and validation logic remain unchanged.

Changes

Cohort / File(s)	Summary
Client-mode allowlist expansion `scripts/nemoclaw-start.sh`	Expanded client-mode condition from exclusively accepting `clientMode == 'webchat'` to also approving `clientMode == 'cli'`, enabling CLI clients to pass the auto-pair watcher validation.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

Poem

A CLI hops in through the gate,
Where webchat once stood tall and great—
Now both modes dance in harmony,
The watcher smiles, "You're welcome, see?" 🐰✨

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately and concisely describes the main change: allowing CLI clients in the auto-pair watcher functionality.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

@sayalinvidia

## Summary Use the OpenShell provider system for messaging credential injection instead of raw env var passthrough. Discord, Slack, and Telegram tokens now flow through the placeholder/proxy pipeline — sandbox processes never see real values. The host-side Telegram bridge is removed; messaging channels are baked into `openclaw.json` at image build time via `NEMOCLAW_MESSAGING_CHANNELS_B64`, and the L7 proxy rewrites placeholders with real secrets at egress — no runtime config patching needed. Signed-off-by: Aaron Erickson <aerickson@nvidia.com> ## Related Issues Fixes #1109 Fixes #616 Fixes #1310 Supersedes #617 ## Changes - **`bin/lib/onboard.js`** — Create `generic` providers for Discord, Slack, and Telegram tokens via `upsertProvider()`. Attach to sandbox via `--provider` flags. Replace individual env var deletes with a comprehensive blocklist. Bake messaging channel config into `openclaw.json` at build time. Collect Telegram user ID for DM allowlisting. - **`Dockerfile`** — Accept `NEMOCLAW_MESSAGING_CHANNELS_B64` build arg and inject channel config into `openclaw.json` at image build time. - **`scripts/nemoclaw-start.sh`** — Remove dead runtime `openclaw.json` patching from `configure_messaging_channels`. Allow CLI clients in auto-pair watcher. - **`nemoclaw/src/lib/services.ts`** — Remove stale `telegram-bridge` spawn. - **`scripts/telegram-bridge.js`** — Removed (replaced by native OpenClaw channels via providers). - **`test/onboard.test.js`** — Verify provider create commands, `--provider` flags on sandbox create, and that real token values never appear in the sandbox create command. - **`test/credential-exposure.test.js`** — Updated for expanded blocklist coverage. - **`test/e2e/messaging-providers.test.sh`** — New E2E test: provider creation, sandbox attachment, DM allowlisting. ## Thanks - @sayalinvidia — tested Discord end-to-end, diagnosed that Landlock makes `openclaw.json` immutable at runtime in non-root mode, and proposed the build-time bake approach via `NEMOCLAW_MESSAGING_CHANNELS_B64` that made this work (PR #1501) - @mercl-lau — found the stale `telegram-bridge` spawn in `services.ts` that silently crashed after the bridge script was removed - @stevenrick — tested Telegram on Brev, independently confirmed the Landlock issue, and found that the auto-pair watcher rejected CLI clients (also opened #1496) ## Type of Change - [x] Code change for a new feature, bug fix, or refactor. - [ ] Code change with doc updates. - [ ] Doc only. Prose changes without code sample modifications. - [ ] Doc only. Includes code sample changes. ## Testing - [x] `npx prek run --all-files` passes (or equivalently `make check`). - [x] `npm test` passes. - [ ] `make docs` builds without warnings. (for doc-only changes) - [x] E2E validated with real bot tokens on Brev instance ## Checklist ### General - [x] I have read and followed the [contributing guide](https://github.com/NVIDIA/NemoClaw/blob/main/CONTRIBUTING.md). - [ ] I have read and followed the [style guide](https://github.com/NVIDIA/NemoClaw/blob/main/docs/CONTRIBUTING.md). (for doc-only changes) ### Code Changes - [x] Formatters applied — `npx prek run --all-files` auto-fixes formatting (or `make format` for targeted runs). - [x] Tests added or updated for new or changed behavior. - [x] No secrets, API keys, or credentials committed. - [ ] Doc pages updated for any user-facing behavior changes (new commands, changed defaults, new features, bug fixes that contradict existing docs). ### Doc Changes N/A --------- Signed-off-by: Aaron Erickson <aerickson@nvidia.com> Co-authored-by: Carlos Villela <cvillela@nvidia.com> Co-authored-by: sayalinvidia <sayalinvidia@users.noreply.github.com>

ericksoa · 2026-04-05T20:56:22Z

Superseded by PR #1081, which cherry picked this into the overall fix on messaging. Thanks for the contribution @stevenrick !

@sayalinvidia

…IA#1081) ## Summary Use the OpenShell provider system for messaging credential injection instead of raw env var passthrough. Discord, Slack, and Telegram tokens now flow through the placeholder/proxy pipeline — sandbox processes never see real values. The host-side Telegram bridge is removed; messaging channels are baked into `openclaw.json` at image build time via `NEMOCLAW_MESSAGING_CHANNELS_B64`, and the L7 proxy rewrites placeholders with real secrets at egress — no runtime config patching needed. Signed-off-by: Aaron Erickson <aerickson@nvidia.com> ## Related Issues Fixes NVIDIA#1109 Fixes NVIDIA#616 Fixes NVIDIA#1310 Supersedes NVIDIA#617 ## Changes - **`bin/lib/onboard.js`** — Create `generic` providers for Discord, Slack, and Telegram tokens via `upsertProvider()`. Attach to sandbox via `--provider` flags. Replace individual env var deletes with a comprehensive blocklist. Bake messaging channel config into `openclaw.json` at build time. Collect Telegram user ID for DM allowlisting. - **`Dockerfile`** — Accept `NEMOCLAW_MESSAGING_CHANNELS_B64` build arg and inject channel config into `openclaw.json` at image build time. - **`scripts/nemoclaw-start.sh`** — Remove dead runtime `openclaw.json` patching from `configure_messaging_channels`. Allow CLI clients in auto-pair watcher. - **`nemoclaw/src/lib/services.ts`** — Remove stale `telegram-bridge` spawn. - **`scripts/telegram-bridge.js`** — Removed (replaced by native OpenClaw channels via providers). - **`test/onboard.test.js`** — Verify provider create commands, `--provider` flags on sandbox create, and that real token values never appear in the sandbox create command. - **`test/credential-exposure.test.js`** — Updated for expanded blocklist coverage. - **`test/e2e/messaging-providers.test.sh`** — New E2E test: provider creation, sandbox attachment, DM allowlisting. ## Thanks - @sayalinvidia — tested Discord end-to-end, diagnosed that Landlock makes `openclaw.json` immutable at runtime in non-root mode, and proposed the build-time bake approach via `NEMOCLAW_MESSAGING_CHANNELS_B64` that made this work (PR NVIDIA#1501) - @mercl-lau — found the stale `telegram-bridge` spawn in `services.ts` that silently crashed after the bridge script was removed - @stevenrick — tested Telegram on Brev, independently confirmed the Landlock issue, and found that the auto-pair watcher rejected CLI clients (also opened NVIDIA#1496) ## Type of Change - [x] Code change for a new feature, bug fix, or refactor. - [ ] Code change with doc updates. - [ ] Doc only. Prose changes without code sample modifications. - [ ] Doc only. Includes code sample changes. ## Testing - [x] `npx prek run --all-files` passes (or equivalently `make check`). - [x] `npm test` passes. - [ ] `make docs` builds without warnings. (for doc-only changes) - [x] E2E validated with real bot tokens on Brev instance ## Checklist ### General - [x] I have read and followed the [contributing guide](https://github.com/NVIDIA/NemoClaw/blob/main/CONTRIBUTING.md). - [ ] I have read and followed the [style guide](https://github.com/NVIDIA/NemoClaw/blob/main/docs/CONTRIBUTING.md). (for doc-only changes) ### Code Changes - [x] Formatters applied — `npx prek run --all-files` auto-fixes formatting (or `make format` for targeted runs). - [x] Tests added or updated for new or changed behavior. - [x] No secrets, API keys, or credentials committed. - [ ] Doc pages updated for any user-facing behavior changes (new commands, changed defaults, new features, bug fixes that contradict existing docs). ### Doc Changes N/A --------- Signed-off-by: Aaron Erickson <aerickson@nvidia.com> Co-authored-by: Carlos Villela <cvillela@nvidia.com> Co-authored-by: sayalinvidia <sayalinvidia@users.noreply.github.com>

@sayalinvidia

…IA#1081) ## Summary Use the OpenShell provider system for messaging credential injection instead of raw env var passthrough. Discord, Slack, and Telegram tokens now flow through the placeholder/proxy pipeline — sandbox processes never see real values. The host-side Telegram bridge is removed; messaging channels are baked into `openclaw.json` at image build time via `NEMOCLAW_MESSAGING_CHANNELS_B64`, and the L7 proxy rewrites placeholders with real secrets at egress — no runtime config patching needed. Signed-off-by: Aaron Erickson <aerickson@nvidia.com> ## Related Issues Fixes NVIDIA#1109 Fixes NVIDIA#616 Fixes NVIDIA#1310 Supersedes NVIDIA#617 ## Changes - **`bin/lib/onboard.js`** — Create `generic` providers for Discord, Slack, and Telegram tokens via `upsertProvider()`. Attach to sandbox via `--provider` flags. Replace individual env var deletes with a comprehensive blocklist. Bake messaging channel config into `openclaw.json` at build time. Collect Telegram user ID for DM allowlisting. - **`Dockerfile`** — Accept `NEMOCLAW_MESSAGING_CHANNELS_B64` build arg and inject channel config into `openclaw.json` at image build time. - **`scripts/nemoclaw-start.sh`** — Remove dead runtime `openclaw.json` patching from `configure_messaging_channels`. Allow CLI clients in auto-pair watcher. - **`nemoclaw/src/lib/services.ts`** — Remove stale `telegram-bridge` spawn. - **`scripts/telegram-bridge.js`** — Removed (replaced by native OpenClaw channels via providers). - **`test/onboard.test.js`** — Verify provider create commands, `--provider` flags on sandbox create, and that real token values never appear in the sandbox create command. - **`test/credential-exposure.test.js`** — Updated for expanded blocklist coverage. - **`test/e2e/messaging-providers.test.sh`** — New E2E test: provider creation, sandbox attachment, DM allowlisting. ## Thanks - @sayalinvidia — tested Discord end-to-end, diagnosed that Landlock makes `openclaw.json` immutable at runtime in non-root mode, and proposed the build-time bake approach via `NEMOCLAW_MESSAGING_CHANNELS_B64` that made this work (PR NVIDIA#1501) - @mercl-lau — found the stale `telegram-bridge` spawn in `services.ts` that silently crashed after the bridge script was removed - @stevenrick — tested Telegram on Brev, independently confirmed the Landlock issue, and found that the auto-pair watcher rejected CLI clients (also opened NVIDIA#1496) ## Type of Change - [x] Code change for a new feature, bug fix, or refactor. - [ ] Code change with doc updates. - [ ] Doc only. Prose changes without code sample modifications. - [ ] Doc only. Includes code sample changes. ## Testing - [x] `npx prek run --all-files` passes (or equivalently `make check`). - [x] `npm test` passes. - [ ] `make docs` builds without warnings. (for doc-only changes) - [x] E2E validated with real bot tokens on Brev instance ## Checklist ### General - [x] I have read and followed the [contributing guide](https://github.com/NVIDIA/NemoClaw/blob/main/CONTRIBUTING.md). - [ ] I have read and followed the [style guide](https://github.com/NVIDIA/NemoClaw/blob/main/docs/CONTRIBUTING.md). (for doc-only changes) ### Code Changes - [x] Formatters applied — `npx prek run --all-files` auto-fixes formatting (or `make format` for targeted runs). - [x] Tests added or updated for new or changed behavior. - [x] No secrets, API keys, or credentials committed. - [ ] Doc pages updated for any user-facing behavior changes (new commands, changed defaults, new features, bug fixes that contradict existing docs). ### Doc Changes N/A --------- Signed-off-by: Aaron Erickson <aerickson@nvidia.com> Co-authored-by: Carlos Villela <cvillela@nvidia.com> Co-authored-by: sayalinvidia <sayalinvidia@users.noreply.github.com>

fix(scripts): allow CLI clients in auto-pair watcher

2326922

Add 'cli' to ALLOWED_MODES so sandbox-internal CLI commands can connect to the gateway. Related: #1310 Signed-off-by: Steven Rick <srick@nvidia.com>

stevenrick mentioned this pull request Apr 5, 2026

fix(security): use providers for messaging credential injection #1081

Merged

14 tasks

Merge branch 'main' into fix/auto-pair-cli-clients

6bd34a2

ericksoa closed this Apr 5, 2026

stevenrick deleted the fix/auto-pair-cli-clients branch May 13, 2026 06:20

wscurran added the bug-fix PR fixes a bug or regression label Jun 8, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(scripts): allow CLI clients in auto-pair watcher#1496

fix(scripts): allow CLI clients in auto-pair watcher#1496
stevenrick wants to merge 2 commits into
NVIDIA:mainfrom
stevenrick:fix/auto-pair-cli-clients

stevenrick commented Apr 5, 2026 •

edited by ericksoa

Loading

Uh oh!

coderabbitai Bot commented Apr 5, 2026 •

edited

Loading

Walkthrough

Changes

Estimated code review effort

Poem

❌ Failed checks (1 warning)

Uh oh!

ericksoa commented Apr 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

stevenrick commented Apr 5, 2026 • edited by ericksoa Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Related Issue

Changes

Type of Change

Testing

Checklist

General

Code Changes

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Apr 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Poem

❌ Failed checks (1 warning)

Uh oh!

ericksoa commented Apr 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

stevenrick commented Apr 5, 2026 •

edited by ericksoa

Loading

coderabbitai Bot commented Apr 5, 2026 •

edited

Loading