fix(local_ai): hard-override to disabled until explicit opt-in (#573)

[oxoxDev]

Summary

• Add opt_in_confirmed marker to local AI config; bootstrap hard-overrides enabled=false unless it is true.
• apply_preset sets the marker on any non-disabled tier (user's explicit choice) and clears it on disabled.
• Onboarding cloud-default presentation preserved; tests cover opt-in and stale-config override paths.
• Settings panel continues to render all 5 tier cards (with non-MVP tiers shown as "Coming soon" / non-selectable per PR feat(local_ai): MVP model lockdown — lock selection to 2-4 GB tier (#573) #588) — roadmap visibility is the intended UX.

Problem

PR #588 (MVP model lockdown) clamped selected_tier to the allowlisted range but did not gate enablement. On upgrade, users who already had local_ai.enabled = true with a saved selected_tier (from the pre-MVP default-on era) would still boot with local AI running, bypassing the explicit opt-in the MVP expects.

Reported behaviour: "onboarding flow didn't come for me, but the model selected was for 2-4 GB one, which shouldn't be the case — it should be default cloud, with only a single model available to opt in."

Closes #573

Solution

Close the enablement gap with a config-schema marker + bootstrap gate, defense-in-depth against stale configs and manual edits:

1. Schema (src/openhuman/config/schema/local_ai.rs): Add opt_in_confirmed: bool with #[serde(default)] → defaults to false for any existing config missing the field.
2. Bootstrap gate (src/openhuman/local_ai/service/bootstrap.rs): config_with_recommended_tier_if_unselected hard-overrides enabled = false whenever opt_in_confirmed == false, regardless of device RAM or selected_tier. Every pre-MVP upgrade path therefore lands on cloud fallback.
3. Opt-in write path (src/openhuman/local_ai/schemas.rs): apply_preset is the single source of truth for the marker — any non-disabled tier flips it to true; disabled clears it. Onboarding and Settings already call apply_preset, so no frontend changes are required.

The Settings preset catalog continues to return all_presets() so non-MVP tiers render as "Coming soon" cards (UX established in PR #588), while the existing RPC guard rejects any apply_preset for a non-MVP tier.

Submission Checklist

• Unit tests — cargo test --lib -- local_ai::service::bootstrap (6/6 pass, including new bootstrap_overrides_stale_selected_tier_without_opt_in and renamed bootstrap_honors_opt_in_on_* cases); yarn test:unit LocalAIStep (6/6 pass); yarn test:unit localAiBootstrap (2/2 pass)
• Integration — cargo test --test json_rpc_e2e json_rpc_local_ai_device_profile_and_presets asserts the full 5-preset catalog shape; this PR keeps that contract.
• Doc comments — /// on the new opt_in_confirmed field explaining migration semantics; module-level rationale on the bootstrap gate
• Inline comments — Bootstrap gate explains why each non-opted-in state (fresh install, pre-MVP upgrade with stale selected_tier, manual config edit) is overridden; apply_preset explains the marker flip as the single source of truth

Impact

• Migration: Existing configs without opt_in_confirmed (pre-feat(local_ai): MVP model lockdown — lock selection to 2-4 GB tier (#573) #588 and post-feat(local_ai): MVP model lockdown — lock selection to 2-4 GB tier (#573) #588 upgrades) deserialise the field as false → next boot lands on cloud fallback with enabled = false. selected_tier is preserved on disk for UX display; nothing is destructively rewritten.
• UX: Users on the affected migration path see cloud-default behaviour and must re-opt-in via onboarding or Settings → Local AI. This matches the MVP product decision that local AI is strictly opt-in.
• Settings UI: Unchanged — all 5 tier cards remain visible, only ram_2_4gb is selectable (per PR feat(local_ai): MVP model lockdown — lock selection to 2-4 GB tier (#573) #588's "Coming soon" lockdown).
• Performance: No runtime change beyond one extra boolean check at bootstrap.
• Platform: Desktop-only (local AI never ran on mobile/web).

Related

• Issue: MVP: auto-provision local model at startup and lock selection to 2–4 GB only #573
• Prior PR: feat(local_ai): MVP model lockdown — lock selection to 2-4 GB tier (#573) #588 — MVP model lockdown clamped tier selection but did not gate enablement. This PR closes the enablement gap that surfaced after feat(local_ai): MVP model lockdown — lock selection to 2-4 GB tier (#573) #588 was merged.

🤖 Generated with Claude Code

[coderabbitai]

Warning

Rate limit exceeded

@oxoxDev has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 28 minutes and 25 seconds before requesting another review.

Your organization is not enrolled in usage-based pricing. Contact your admin to enable usage-based pricing to continue reviews beyond the rate limit, or try again in 28 minutes and 25 seconds.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 49d942b8-12a4-496e-b401-57f72a468a56

📥 Commits

Reviewing files that changed from the base of the PR and between 5082f3e and c81291a.

📒 Files selected for processing (5)

• app/src/pages/onboarding/steps/LocalAIStep.tsx
• app/src/pages/onboarding/steps/__tests__/LocalAIStep.test.tsx
• src/openhuman/config/schema/local_ai.rs
• src/openhuman/local_ai/schemas.rs
• src/openhuman/local_ai/service/bootstrap.rs

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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