Repository navigation
docs(reference): document --dangerously-skip-permissions onboard flag - #2223
Conversation
The --dangerously-skip-permissions flag is accepted by the onboard command (src/lib/onboard-command.ts, src/lib/onboard.ts) and shown in the CLI usage line (src/lib/onboard-command.ts:39) but has no entry in the reference docs, so users have no context on what the flag disables, when it is intended for use, or the security implications. Add a subsection under `nemoclaw onboard` (after --from) that: - Leads with a warning that it is development-only and unsafe for production credentials or shared hosts. - Enumerates the concrete effects: network policy (all endpoints open), filesystem (sandbox home writable), and what stays gated (messaging and inference credentials). - Covers both the flag form and the NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1 env var form for non-interactive runs. - Notes that the flag is persisted on the registry entry so `nemoclaw <sandbox> status` surfaces it, and that re-running onboard without the flag restores the default policy. - Adds the flag to the top-level usage line for completeness. Signed-off-by: latenighthackathon <latenighthackathon@users.noreply.github.com>
📝 WalkthroughWalkthroughDocumentation update to the Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~3 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
There was a problem hiding this comment.
🧹 Nitpick comments (1)
docs/reference/commands.md (1)
180-200: Align this section with docs style: one sentence per line and active voice.Please split multi-sentence lines (Lines 180, 193, and 199) into one sentence per source line, and rewrite passive constructions like “is honored” / “is persisted” into active voice.
As per coding guidelines, "One sentence per line in source (makes diffs readable). Flag paragraphs where multiple sentences appear on the same line." and "Active voice required. Flag passive constructions."
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@docs/reference/commands.md` around lines 180 - 200, Split any source lines that contain multiple sentences in the paragraphs describing the --dangerously-skip-permissions flag, the NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS env var usage, and the sandbox registry persistence so each sentence lives on its own line; also convert passive phrasing to active voice (e.g., change "is honored via NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1" to "nemoclaw honors NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1" and change "is persisted on the sandbox registry entry" to "nemoclaw persists the flag on the sandbox registry entry"), ensure the warning/logging sentence is a single active-voice sentence, and keep the command examples (--dangerously-skip-permissions and the env-var onboarding invocation) unchanged.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@docs/reference/commands.md`:
- Around line 180-200: Split any source lines that contain multiple sentences in
the paragraphs describing the --dangerously-skip-permissions flag, the
NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS env var usage, and the sandbox registry
persistence so each sentence lives on its own line; also convert passive
phrasing to active voice (e.g., change "is honored via
NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1" to "nemoclaw honors
NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1" and change "is persisted on the sandbox
registry entry" to "nemoclaw persists the flag on the sandbox registry entry"),
ensure the warning/logging sentence is a single active-voice sentence, and keep
the command examples (--dangerously-skip-permissions and the env-var onboarding
invocation) unchanged.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: 5e28eeb3-b340-4240-9c4f-830e558c72d9
📒 Files selected for processing (1)
docs/reference/commands.md
|
✨ Thanks for submitting this PR that proposes a way to improve the documentation — this could help provide a more accurate and informative resource for users. |
jyaunches
left a comment
There was a problem hiding this comment.
Docs-only PR documenting an existing, well-tested feature. Verified the --dangerously-skip-permissions flag implementation across onboard-command.ts, policies.ts, shields.ts, nemoclaw.ts, and registry.ts. The claim that re-running nemoclaw onboard without the flag tightens the sandbox is accurate — registerSandbox() fully replaces the entry, clearing the flag and rebuilding with the restrictive base policy. Clean to merge.
Summary
The
--dangerously-skip-permissionsflag is accepted bynemoclaw onboard(usage line insrc/lib/onboard-command.ts:39, parse insrc/lib/onboard.ts:6301) and is surfaced onnemoclaw <sandbox> statusfor sandboxes created with it — but it has no entry in the reference docs. A user readingdocs/reference/commands.mdhas no way to learn what the flag disables, when it is intended for use, or the security implications.Changes
Add a subsection under
nemoclaw onboard(after--from) that:NEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1env var form for non-interactive runs.nemoclaw <sandbox> status(Permissions: dangerously-skip-permissions (shields permanently down)), and that re-runningnemoclaw onboardwithout the flag restores the default policy.Testing
Docs-only. Verified:
nemoclaw-blueprint/policies/openclaw-sandbox-permissive.yaml) exists and is used when the flag is set —src/lib/onboard.ts:3640.src/lib/onboard.ts:6302.src/nemoclaw.ts:1443.Checklist
Signed-off-by: latenighthackathon latenighthackathon@users.noreply.github.com
Summary by CodeRabbit
Documentation
--dangerously-skip-permissionsflag fornemoclaw onboardcommand, disabling network and filesystem restrictions in development/testing environmentsNEMOCLAW_DANGEROUSLY_SKIP_PERMISSIONS=1environment variable for non-interactive setups