docs/inference/switch-inference-providers.md (1)

149-149: ⚡ Quick win

Format default value as inline code.

Line 149 uses unset as a parameter value but does not wrap it in inline code formatting.

Proposed fix

-| `NEMOCLAW_AGENT_HEARTBEAT_EVERY` | Go-style duration (`30m`, `1h`, `0m` to disable) | unset (OpenClaw default) |
+| `NEMOCLAW_AGENT_HEARTBEAT_EVERY` | Go-style duration (`30m`, `1h`, `0m` to disable) | `unset` (OpenClaw default) |

As per coding guidelines "CLI commands, file paths, flags, parameter names, and values must use inline code formatting."

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/inference/switch-inference-providers.md` at line 149, The table row for
the parameter NEMOCLAW_AGENT_HEARTBEAT_EVERY uses the plain word unset for the
default value; update that cell so the default value is formatted as inline code
(e.g., `unset`) to follow the guideline that CLI flags, parameter names and
values use inline code formatting and match the surrounding table style.

test/generate-openclaw-config.test.ts (1)

246-264: ⚡ Quick win

Add explicit empty-string heartbeat test for build-arg parity.

Line 246 validates the unset case, but the Docker build path commonly passes an empty string. Add one assertion to lock in that "" also omits agents.defaults.heartbeat.

Proposed test addition

   it("omits heartbeat when NEMOCLAW_AGENT_HEARTBEAT_EVERY is unset", () => {
     const config = runConfigScript();
     expect(config.agents.defaults.heartbeat).toBeUndefined();
   });

+  it("omits heartbeat when NEMOCLAW_AGENT_HEARTBEAT_EVERY is empty", () => {
+    const config = runConfigScript({ NEMOCLAW_AGENT_HEARTBEAT_EVERY: "" });
+    expect(config.agents.defaults.heartbeat).toBeUndefined();
+  });
+
   it("propagates heartbeat cadence into agents.defaults.heartbeat.every", () => {

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@test/generate-openclaw-config.test.ts` around lines 246 - 264, Add a new test
that covers the empty-string build-arg case by calling runConfigScript({
NEMOCLAW_AGENT_HEARTBEAT_EVERY: "" }) and asserting that
config.agents.defaults.heartbeat is undefined; place it alongside the existing
heartbeat tests (the ones using "unset", "30m", "0m", and malformed values) so
the suite verifies that an empty string omits the heartbeat field just like the
unset case.

src/lib/onboard.ts (1)

2364-2535: Run the onboarding E2E suite for this src/lib/onboard.ts change path.

Because this touches core sandbox build/config wiring, it’s worth running the recommended nightly E2E job set before merge to catch integration regressions.

As per coding guidelines: src/lib/onboard.ts: “This file contains core onboarding logic. Changes here affect the full sandbox creation and configuration flow.” and the listed E2E recommendations.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/lib/onboard.ts` around lines 2364 - 2535, The change modifies sandbox
image build metadata in patchStagedDockerfile (ARG replacements including
BASE_IMAGE, NEMOCLAW_* args, proxy, messaging and web search flags); before
merging, run the onboarding E2E suite (the recommended nightly job set) to
validate full sandbox creation: build the image, confirm BASE_IMAGE pinning
behavior, ensure NEMOCLAW_* ARG values (MODEL, PROVIDER_KEY, PRIMARY_MODEL_REF,
INFERENCE_* and BUILD_ID) propagate, verify proxy env handling
(NEMOCLAW_PROXY_HOST/PORT), messaging/telegram/discord arg encoding, web-search
toggle, and that the resulting sandbox boots and exposes the dashboard without
device auth.

Dockerfile (1)

265-270: Run the Dockerfile E2E matrix before merge.

Since this changes sandbox image build-time config behavior, run the recommended E2Es (cloud-e2e, sandbox-survival-e2e, hermes-e2e, rebuild-openclaw-e2e) to validate onboard/rebuild behavior end-to-end.

As per coding guidelines, Dockerfile: This file affects the sandbox container image. Layer ordering, permissions, and baked config changes are only testable with a real container build. E2E test recommendation: cloud-e2e, sandbox-survival-e2e, hermes-e2e, rebuild-openclaw-e2e.

Also applies to: 323-323

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@Dockerfile` around lines 265 - 270, This change introduces a build-time ARG
(NEMOCLAW_AGENT_HEARTBEAT_EVERY) in the Dockerfile that affects sandbox image
behavior; before merging, run the full Dockerfile E2E matrix to validate
rebuild/onboard behavior: execute cloud-e2e, sandbox-survival-e2e, hermes-e2e,
and rebuild-openclaw-e2e against an image built with the new ARG and verify
openclaw.json behavior and heartbeat cadence; report any failures and only merge
once all four E2Es pass.

docs/inference/switch-inference-providers.md (1)

170-175: 💤 Low value

One sentence per line — formatting rule violated.

Sentence 1 ends and sentence 2 begins on the same physical line (line 170). The style guide requires one sentence per source line for readable diffs.

📝 Suggested reformatting

-`NEMOCLAW_AGENT_HEARTBEAT_EVERY` sets `agents.defaults.heartbeat.every`. Set it
-to `0m` to disable the periodic heartbeat when it disrupts long-running agent
-turns; leave it unset to preserve the OpenClaw default cadence. `openclaw.json`
-is immutable at runtime, so the in-sandbox `openclaw config set` command cannot
-change this — rebuild the sandbox via `nemoclaw onboard --resume` to apply a
-new value.
+`NEMOCLAW_AGENT_HEARTBEAT_EVERY` sets `agents.defaults.heartbeat.every`.
+Set it to `0m` to disable the periodic heartbeat when it disrupts long-running agent turns; leave it unset to preserve the OpenClaw default cadence.
+`openclaw.json` is immutable at runtime, so the in-sandbox `openclaw config set` command cannot change this — rebuild the sandbox via `nemoclaw onboard --resume` to apply a new value.

As per coding guidelines: "One sentence per line in source (makes diffs readable). Flag paragraphs where multiple sentences appear on the same line."

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/inference/switch-inference-providers.md` around lines 170 - 175, The
paragraph contains multiple sentences on the same physical line; split it so
each sentence is on its own source line to satisfy the "one sentence per line"
guideline. Specifically, break the line that mentions
NEMOCLAW_AGENT_HEARTBEAT_EVERY, agents.defaults.heartbeat.every, the `0m`
disable note, the immutability of openclaw.json, the inability of the in-sandbox
`openclaw config set` command to change it, and the instruction to rebuild the
sandbox via `nemoclaw onboard --resume` into separate lines so each sentence
stands alone; keep the same wording and the same references to
NEMOCLAW_AGENT_HEARTBEAT_EVERY, agents.defaults.heartbeat.every, openclaw.json,
`openclaw config set`, and `nemoclaw onboard --resume`.

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes