MCA v0.2.0-alpha

MCA v0.2.0-alpha

Feature and compatibility update for the public alpha line.

Highlights

• Added configurable web search with direct public URL reading, keyword search,
  source cards, search traces, local diagnostics, and per-turn controls from the
  chat composer.
• Added research-mode controls that follow the same lightweight capsule pattern
  as reasoning mode.
• Added search provider setup for SearxNG, Brave Search, Tavily, Jina Search,
  and custom JSON search gateways, with in-app guidance for obtaining endpoints
  or API keys.
• Added the in-app API Usage Guide for OpenAI-compatible local API setup,
  trusted LAN access, supported paths, JSON replies, and SSE streaming.
• Improved local API compatibility for third-party OpenAI-compatible clients by
  preserving client-provided system messages and character-card prompts.
• Added authenticated local API vision self-test and /v1/models vision
  diagnostics without exposing generated local API keys in logs or shell
  commands.
• Improved cloud multimodal request forwarding for compatible OpenAI-style and
  Anthropic-style vision engines by preserving inline image data URLs.
• Added experimental local vision support for multimodal GGUF models with
  matching mmproj / projector files.
• Added a ModelScope recommendation bundle for MiniCPM-V 4.6 Q4 local vision:
  MCA downloads the main GGUF and matching projector, then registers and binds
  them together.
• Added local model-card actions to bind or replace mmproj / projector files,
  including projector files downloaded from remote model file lists.
• Reworked cloud inference-engine and image-generation-engine setup from small
  dialogs into full mobile pages with clearer forms and visible test results.
• Kept cloud inference engines and image-generation engines separated, including
  support for OpenAI-compatible chat, Anthropic Messages, OpenAI Images,
  DashScope Image, and custom image paths.
• Improved model-management cards so action buttons no longer cover model
  names, while local image engines and local vision bundles remain clearly
  labeled.
• Added assistant and role-card improvements, including MCA role-card
  import/export and compatibility import for common nested character-card
  data fields.
• Preserved the existing MCA chat surface and visual language while aligning
  secondary-page transitions and cleaning up menu text.

Local API Notes

Use http://127.0.0.1:11435/v1 for same-device clients.

Use http://<phone-lan-ip>:11435/v1 for another trusted device on the same
network, after enabling open port inside MCA.

Supported paths:

• GET /health
• GET /v1/models
• POST /v1/chat/completions
• GET / for the built-in web chat page with text and image attachment input

/v1/models includes MCA extension fields such as vision_ready and
vision_projector so clients can detect whether the loaded local model is ready
for image input.

The local API key is generated inside MCA and should not be shared in issues,
screenshots, videos, or release notes.

Validation

• :core:download:testDebugUnitTest
• :core:modelstore:testDebugUnitTest
• :core:engine:testDebugUnitTest
• :api:local:testDebugUnitTest
• :app:testDebugUnitTest
• :feature:modelhub:compileDebugKotlin
• :feature:chat:compileDebugKotlin
• :app:compileDebugKotlin
• :app:assembleRelease
• Release APK install and launch smoke test on a real Android device.
• Cloud vision mock smoke test with OpenAI-compatible image input enabled.
• Authenticated local API vision self-test on a real Android device with
  MiniCPM-V 4.6 Q4 and mmproj-model-f16.gguf.
• Model management smoke test covering local models, cloud engines,
  recommendations, and local vision projector setup surfaces.

Known Limits

• MCA does not include model weights, provider accounts, cloud endpoints, or API
  keys.
• Local API chat requires a local model to be loaded first.
• Same-LAN local API access should only be enabled on trusted networks.
• Cloud provider compatibility depends on each provider's exact protocol,
  route, model name, and image-input support.
• Local vision is not automatic for ordinary GGUF files. It requires a
  compatible multimodal GGUF model and matching projector, and the native runner
  must report visionReady=true.
• Local image generation remains experimental. Performance depends heavily on
  device, model bundle, resolution, step count, thermal state, and available
  memory.
• Keyword web search requires a configured search provider. The public JSON
  self-check source is only a protocol check, not a production search engine.

MCA v0.1.0-alpha.3

MCA v0.1.0-alpha.3

Local API persona hotfix for third-party OpenAI-compatible clients.

Highlights

• Fixed local API requests losing the current MCA persona/system prompt when a
  client does not send its own system message.
• Preserved third-party character-card system messages so client-provided
  personas are not overwritten by MCA defaults.
• Persisted generation parameters across app/model reloads, including system
  prompt, sampling settings, context size, thread count, and reasoning mode.
• Unified the AIDL and HTTP local API request parser so both paths share the
  same OpenAI-compatible behavior.
• Added unit tests for persona inheritance, third-party character-card
  precedence, and generation-parameter round trips.

Validation

• :api:local:testDebugUnitTest
• :core:engine:testDebugUnitTest
• :app:assembleDebug
• Device smoke test with OpenAI-compatible /v1/models and
  /v1/chat/completions.
• Verified both non-streaming JSON and stream=true SSE responses preserve the
  third-party character-card system message.

Known Limits

• A local chat model still must be loaded before local API chat requests can
  complete.
• Local image generation remains experimental.

MCA v0.1.0-alpha.2

MCA v0.1.0-alpha.2

Local API compatibility update for the public alpha line.

Highlights

• Verified OpenAI-compatible local API use with a third-party Android client.
• Improved /v1/chat/completions compatibility for common connection tests:
  multiple system messages are normalized and probes without a user turn no
  longer fail local chat templates.
• Improved streaming detection for stream=true, string/number stream flags,
  and Accept: text/event-stream.
• Confirmed SSE responses with data: {...} chunks and final data: [DONE].
• Added foreground keep-alive support for the local API server while it is
  enabled.
• Added cloud multimodal chat input support by sending image attachments to
  OpenAI-compatible and Anthropic-compatible vision models.
• Added experimental local vision support for llama.cpp multimodal GGUF models
  that have a matching mmproj / vision projector file.
• Expanded the in-app API Usage Guide with Base URL guidance, supported paths,
  stream behavior, error explanations, and trusted-network safety notes.
• Updated GitHub documentation for local API setup and compatibility.

Local API Notes

Use http://127.0.0.1:11435/v1 for same-device clients.

Use http://<phone-lan-ip>:11435/v1 for another trusted device on the same
network, after enabling open port inside MCA.

Supported paths:

• GET /health
• GET /v1/models
• POST /v1/chat/completions
• GET /

The API key is generated inside MCA and should not be shared in issues,
screenshots, or release notes.

Known Limits

• A local chat model must be loaded before chat requests can complete.
• Same-LAN open port should only be used on trusted networks.
• Local vision requires a matching multimodal GGUF model and projector file;
  text-only GGUF models cannot identify images.
• Local image generation remains experimental and should not be presented as a
  stable fast phone-side feature yet.

MCA v0.1.0-alpha.1

MCA v0.1.0-alpha.1

First installable alpha package for MCA.

Highlights

• Local-first Android AI workspace.
• Local GGUF chat through the native llama.cpp bridge.
• Cloud chat engines through OpenAI-compatible and Anthropic Messages protocols.
• Cloud image engines through OpenAI Images, DashScope Image, and custom paths.
• Experimental local image generation through stable-diffusion.cpp.
• ModelScope-oriented recommendations and resumable downloads.
• Agent diagnostics, local benchmarks, and parameter recommendations.

Known Limits

• Local image generation is experimental and requires complete model bundles.
  Treat it as an advanced test path, not a guaranteed stable phone-side feature.
• Large local models depend heavily on device memory, storage speed, and thermal state.
• Cloud engines require user-provided API keys and provider endpoints.
• This alpha APK is built for arm64-v8a Android devices.

See docs/MODEL_COMPATIBILITY.md for protocol and model-bundle expectations.

Verification

• Signed arm64-v8a release APK clean-install verified on 2026-06-23.
• Test device: 25091RP04C, Android 16 / SDK 36, SoC property SM8750P.
• Smoke record: docs/releases/v0.1.0-alpha.1-smoke.md.

Install

Download the APK from this release and install it manually on an Android device.
Android may require allowing installation from your browser or file manager.