[ucguy4u]

Summary

Mind can now run a short tool round with a LiteRT-LM package that supports function calling. The model may ask for the current time, a calendar read, or a LifeTrack status. Airo executes that lookup and sends the result back into the same conversation.

• The Android plugin registers those tools with LiteRT and turns automatic tool execution off, so the native runtime cannot perform the action itself.
• Replies now include tool-call names and arguments. A follow-up turn sends the tool result.
• Tools outside that lookup list are refused. Flashlight, mail, contacts, and other side effects stay on the existing skill path.
• Packages that do not support function calling keep the token stream from feat(ai): stream LiteRT-LM replies and read ModelInfo #2116.

Test Plan

• flutter test in packages/core_ai for the tool loop, LiteRT adapter, execution adapter, and router
• flutter test in app for the LiteRT service, warmup, and receipt extraction
• flutter test test/agent_chat/data/services/assistant_runtime_service_test.dart --name LiteRT in packages/feature_mind
• dart analyze on the changed Dart libraries
• scripts/check-docs-completeness.sh --base origin/main --head HEAD
• Manual device verification documented, or not applicable

Verification environment: Host-only. No Android device in this environment, so the Kotlin tool channel was not exercised on hardware.

Agent Policy

• Linked issue includes a Critical Agent Gate.
• Linked issue includes a Feature Packet or documents why one is not needed.
• Cross-agent contract is documented for framework/application boundary work.
• Deterministic use cases and automation flows are linked or included.
• AI/tool/memory/routine changes include eval, trace, and redaction coverage.
• Android Emulator was not used unless the linked issue explicitly accepted AIRO_ALLOW_ANDROID_EMULATOR=true.

Follow-up to merged #2116. Framework owns the LiteRT tool channel. Mind owns which lookup tools are offered and how chat runs them.

Council Review

• I checked the Decision Matrix and listed every required role above.
• Every package touched has a module.yaml; if not, this PR adds one (scripts/check-module-manifests.py enforces this in CI).
• owner/reviewers in any touched module.yaml still match who actually reviewed this PR — updated if the roster has drifted.

User-Facing Documentation

• I updated docs/wiki for any user-visible feature, route, platform, install, AI/model, privacy, file-type, media, game, finance, or troubleshooting change.
• No docs/wiki update is needed because this PR has no user-visible behavior or documentation impact.

Plugin and Size Impact

• This PR adds new dependencies
• This PR adds or grows app/packages/plugins code
• Size impact is documented below
• Any module over 3 MB is a plugin or has an explicit plugin marker
• Any plugin over 5 MB has cache-management documentation

Size impact / plugin justification:

No new dependency. litertlm-android stays pinned at 0.18.0. The change is the existing plugin's tool-call bridge.

Checklist

• Code is formatted.
• Tests or manual verification are included above.
• Sensitive data, credentials, and signing artifacts are not committed.

Release Notes

• User-facing change documented, or not applicable

  

[github-actions]

Plugin Module Size Gate

Policy: modules over 3 MB must be delivered as plugins; plugin modules over 5 MB must document cache management.

Module	Size	Type	Status
packages/core_ai	0.58 MB	bundled	OK
packages/feature_mind	2.94 MB	bundled	WARN: 1-3 MB requires PR size justification

[github-actions]

🚀 PR Quick Check Summary

Check	Status	Description
PR Validation	❌ failure	Title format, docs, bundled model guardrail
Code Quality	✅ success	Analyze, formatting
Core Tests	✅ success	Core package unit tests
Airo Mind Shell	✅ success	Mind flavour analyze, tests, entrypoint compile

💡 Note: Full app tests, coverage reports, and security scans run on merge to main.

View Details