feat(app): ✨ adding Preferences/Wingman for ollama, OpenRouter and OpenAI by dvorka · Pull Request #1558 · dvorka-oss/mindforger

dvorka · 2026-02-08T19:20:58Z

Tasks:

green Linux build
green Linux gtest based tests
green macOS build
green Win build

This PR brings:

ollama support & configuration.
OpenRouter support & configuration.
rewritten OpenAI configuration.

augmentcode · 2026-02-08T19:26:52Z

🤖 Augment PR Summary

Summary: Adds a new “Wingman2” Preferences tab for managing multiple LLM provider profiles (OpenAI/ollama).
Changes:

Registers new dialog sources/headers in app/app.pro.
Implements Wingman2Tab UI with provider dropdown, details panel, and Add/Edit/Remove/Test actions.
Adds dialogs to select a provider type and configure OpenAI/ollama settings (API key/URL + model selection).
Extends Configuration with LlmProviderConfig, provider collection APIs, active-provider tracking, and a legacy-migration helper.
Fixes OllamaWingman::listModels() to store the parsed model name.
Enhances OpenAiWingman::listModels() with an HTTP GET to /v1/models and fallback defaults.
Adds a detailed design doc for Wingman2 LLM configuration.

Technical Notes: Provider “probe” routines currently do basic validation and mark configs as valid/invalid for UI feedback.

_{🤖 Was this summary useful? React with 👍 or 👎}

augmentcode

Review completed. 7 suggestions posted.

Comment augment review to trigger a new review at any time.

augmentcode · 2026-02-08T19:26:52Z

+
+#include <QtWidgets>
+
+#include "../../lib/src/config/configuration.h"


The relative include ../../lib/src/config/configuration.h resolves to app/src/lib/... from this header, which doesn’t exist in the repo, so these new dialogs likely won’t compile. Same issue also appears in the OpenAI/ollama config dialog headers.

Severity: high

Other Locations

app/src/qt/dialogs/openai_config_dialog.h:25

app/src/qt/dialogs/openai_config_dialog.h:26

app/src/qt/dialogs/ollama_config_dialog.h:25

app/src/qt/dialogs/ollama_config_dialog.h:26

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

augmentcode · 2026-02-08T19:26:52Z

+    QPushButton* addProviderButton;
+
+    QGroupBox* providerDetailsGroup;
+    QLabel* providerTypeLabel;


Wingman2Tab declares providerTypeLabel/modelLabel/statusLabel members, but the .cpp constructs local QLabel*s instead; these members remain uninitialized. This is easy to accidentally dereference later and crash.

Severity: low

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

augmentcode · 2026-02-08T19:26:53Z

+    : QWidget(parent),
+      config(Configuration::getInstance())
+{
+    helpLabel = new QLabel(


The help text says Wingman2 config is “to be used by Wingman”, but the current runtime initialization path still appears to read only the legacy wingmanProvider/wingmanOpenAi*/wingmanOllama* fields. This may make the new tab misleading if Wingman doesn’t actually consume llmProviders yet.

Severity: medium

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

augmentcode · 2026-02-08T19:26:53Z

+    // populate providers combo
+    llmProvidersCombo->clear();
+
+    vector<LlmProviderConfig>& providers = config.getLlmProviders();


Wingman2Tab::refresh() populates from config.getLlmProviders() directly; since migrateFromLegacyWingmanConfig() isn’t invoked here, existing Wingman users may see an empty provider list even if legacy config is set.

Severity: medium

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

augmentcode · 2026-02-08T19:26:53Z

+        if (providerType == WINGMAN_PROVIDER_OPENAI) {
+            OpenAiConfigDialog configDialog(this);
+            if (configDialog.exec() == QDialog::Accepted) {
+                config.addLlmProvider(configDialog.getProviderConfig());


Providers added via config.addLlmProvider() look to be stored only in-memory; MarkdownConfigurationRepresentation currently persists only the legacy Wingman fields. That means Wingman2 providers/activeLlmProviderId may be lost after restart.

Severity: high

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

Copilot

Pull request overview

Adds a new “Wingman2” preferences workflow to manage multiple LLM provider configurations (OpenAI + ollama), alongside groundwork in Configuration for storing providers, plus improvements to model discovery and a small ollama bug fix.

Changes:

Introduces Wingman2 UI (tab + dialogs) to add/configure LLM providers and select an active provider.
Extends Configuration with a LlmProviderConfig model and basic CRUD/probe/migration APIs.
Enhances OpenAI model listing via /v1/models and fixes an ollama model-list parsing bug.

Reviewed changes

Copilot reviewed 19 out of 19 changed files in this pull request and generated 22 comments.

Show a summary per file

File	Description
vibe/designs/wingman2-llm-configuration.md	Design spec for Wingman2 provider configuration UX, storage, and probing.
lib/src/mind/ai/llm/openai_wingman.h	Adds declaration for HTTP model listing helper.
lib/src/mind/ai/llm/openai_wingman.cpp	Implements OpenAI `/v1/models` model fetching and fallback defaults.
lib/src/mind/ai/llm/ollama_wingman.cpp	Fixes model name push bug in model listing.
lib/src/config/configuration.h	Adds Wingman2 provider config struct, defaults, and new public APIs.
lib/src/config/configuration.cpp	Implements provider CRUD, active selection, probe stubs, and migration helper.
app/src/qt/dialogs/configuration_dialog.h	Adds `Wingman2Tab` declaration and members.
app/src/qt/dialogs/configuration_dialog.cpp	Adds Wingman2 tab UI/handlers and wires it into Preferences show/save.
app/src/qt/dialogs/openai_config_dialog.{h,cpp}	New OpenAI provider dialog (API key + model + refresh/probe/add).
app/src/qt/dialogs/ollama_config_dialog.{h,cpp}	New ollama provider dialog (URL + model + refresh/probe/add).
app/src/qt/dialogs/add_llm_provider_dialog.{h,cpp}	New dialog to choose provider type before configuring.
app/app.pro	Registers the new dialog sources/headers with qmake.
lib/src/app_info.h	Updates legal copyright year range.
build/ubuntu/debian/copyright	Updates Debian packaging copyright year range.
build/debian/debian/copyright	Updates Debian packaging copyright year range.
.github/copilot-instructions.md	Adds repository-specific Copilot guidance.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-08T19:29:51Z

+    MF_DEBUG(
+        "OpenAiWingman::listModelsHttpGet() parsed response:" << endl
+        << ">>>"
+        << httpResponseJson.dump(4)
+        << "<<<"
+        << endl);


listModelsHttpGet() logs the full parsed JSON response (httpResponseJson.dump(4)). The /v1/models payload can be large, which may bloat logs and add noticeable overhead. Consider logging only a summary (e.g., model count / first N IDs) or gating the full dump behind a more verbose debug flag.

Suggested change

MF_DEBUG(

"OpenAiWingman::listModelsHttpGet() parsed response:" << endl

<< ">>>"

<< httpResponseJson.dump(4)

<< "<<<"

<< endl);

std::size_t modelsCount = 0;

if (httpResponseJson.contains("data") && httpResponseJson["data"].is_array()) {

modelsCount = httpResponseJson["data"].size();

}

MF_DEBUG(

"OpenAiWingman::listModelsHttpGet() parsed response: "

<< "data array size = " << modelsCount << endl);

Copilot · 2026-02-08T19:29:51Z

+    AddLlmProviderDialog(const AddLlmProviderDialog&&) = delete;
+    AddLlmProviderDialog& operator=(const AddLlmProviderDialog&) = delete;
+    AddLlmProviderDialog& operator=(const AddLlmProviderDialog&&) = delete;


AddLlmProviderDialog also declares deleted move operations as const AddLlmProviderDialog&& / operator=(const AddLlmProviderDialog&&). This isn’t the normal move signature and can be misleading; use AddLlmProviderDialog&& (or omit move deletions if not needed).

Suggested change

AddLlmProviderDialog(const AddLlmProviderDialog&&) = delete;

AddLlmProviderDialog& operator=(const AddLlmProviderDialog&) = delete;

AddLlmProviderDialog& operator=(const AddLlmProviderDialog&&) = delete;

AddLlmProviderDialog(AddLlmProviderDialog&&) = delete;

AddLlmProviderDialog& operator=(const AddLlmProviderDialog&) = delete;

AddLlmProviderDialog& operator=(AddLlmProviderDialog&&) = delete;

Copilot · 2026-02-08T19:29:52Z

+    string errorMessage;
+    bool success = false;
+
+    if (provider->providerType == WINGMAN_PROVIDER_OPENAI) {
+        success = config.probeOpenAiProvider(
+            provider->apiKey, provider->llmModel, errorMessage);
+    } else if (provider->providerType == WINGMAN_PROVIDER_OLLAMA) {
+        success = config.probeOllamaProvider(
+            provider->url, provider->llmModel, errorMessage);
+    }
+
+    if (success) {
+        QMessageBox::information(
+            this,
+            tr("Connection Test"),
+            tr("Provider configuration is valid."));
+    } else {
+        QMessageBox::critical(
+            this,
+            tr("Connection Test"),
+            tr("Provider configuration test failed: %1")
+            .arg(QString::fromStdString(errorMessage)));
+    }


handleTestConnection() reports success/failure but never updates provider->isValid, so the status in the “Selected Provider Details” panel won’t reflect the latest test result. Consider setting provider->isValid = success (and persisting via updateLlmProvider if needed) and then refreshing the UI.

Copilot · 2026-02-08T19:29:55Z

+    // generate unique ID using timestamp
+    auto now = chrono::system_clock::now();
+    auto timestamp = chrono::duration_cast<chrono::seconds>(now.time_since_epoch()).count();
+
+    // extract host from URL for display name
+    string host = url;
+    size_t pos = url.find("://");
+    if (pos != string::npos) {
+        host = url.substr(pos + 3);
+    }
+
+    providerConfig.id = "ollama-" + to_string(timestamp);
+    providerConfig.displayName = "ollama " + model + " @ " + host;
+    providerConfig.providerType = WINGMAN_PROVIDER_OLLAMA;
+    providerConfig.url = url;
+    providerConfig.llmModel = model;
+    providerConfig.isValid = configValid;


Provider IDs are generated using seconds-resolution timestamps (e.g., ollama-<seconds>), which can collide if the user adds providers quickly. Since IDs are used as stable keys for update/remove/active selection, consider generating IDs with higher entropy (ms + random/UUID) or checking for collisions and regenerating.

Copilot · 2026-02-08T19:29:55Z

+    return getLlmProviderById(activeLlmProviderId);
+}
+
+void Configuration::addLlmProvider(const LlmProviderConfig& provider) {


addLlmProvider() blindly appends the provider without enforcing unique IDs (despite id being documented as unique). This can create duplicate IDs (e.g., two providers added within the same second using timestamp IDs) and then getLlmProviderById()/setActiveLlmProvider() become ambiguous. Consider rejecting duplicates (and surfacing an error) or auto-renaming/regenerating the ID on collision.

Suggested change

void Configuration::addLlmProvider(const LlmProviderConfig& provider) {

void Configuration::addLlmProvider(const LlmProviderConfig& provider) {

// Enforce unique provider IDs to avoid ambiguous lookups.

if (!provider.id.empty() && getLlmProviderById(provider.id) != nullptr) {

MF_DEBUG("Configuration::addLlmProvider() duplicate ID, provider not added: " << provider.id << endl);

return;

}

Copilot · 2026-02-08T19:29:56Z

+    MF_DEBUG("Configuration::removeLlmProvider() removed: " << id << endl);
+}
+
+void Configuration::setActiveLlmProvider(const string& id) {


setActiveLlmProvider() sets activeLlmProviderId without verifying the ID exists in llmProviders. This can leave the config pointing at a non-existent provider (e.g., after update/remove sequences or malformed persisted data). Consider validating the ID (no-op with error/log, or clear active selection) to keep getActiveLlmProvider() consistent.

Suggested change

void Configuration::setActiveLlmProvider(const string& id) {

void Configuration::setActiveLlmProvider(const string& id) {

// Only set active provider if the ID is known; keep configuration consistent.

if (!id.empty() && getLlmProviderById(id) == nullptr) {

MF_DEBUG("Configuration::setActiveLlmProvider() attempted to set unknown provider id: " << id << endl);

return;

}

…oice-ii-vibe

- Remove legacy per-provider config fields (wingmanOpenAiApiKey, etc.) and replace with std::vector<LlmProviderConfig> + activeLlmProviderId - OpenAiWingman/OpenRouterWingman now take explicit apiKey in constructor - Configuration dialog consolidated from 4 Wingman tabs to 1 WingmanTab - Config serializer/parser updated with backward-compatible migration - mind.cpp initWingman()/getWingman() updated to use new provider API - openai/ollama/openrouter config dialogs fixed to use new API Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 27 out of 28 changed files in this pull request and generated 17 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

    } else {
        // check whether possible
-        if(config.canWingmanOllama()) {
+        if(!config.getLlmProviders().empty()) {


+    // generate unique ID using timestamp
+    auto now = chrono::system_clock::now();
+    auto timestamp = chrono::duration_cast<chrono::seconds>(now.time_since_epoch()).count();
+    providerConfig.id = "openai-" + to_string(timestamp);


+    // generate unique ID using timestamp
+    auto now = chrono::system_clock::now();
+    auto timestamp = chrono::duration_cast<chrono::seconds>(now.time_since_epoch()).count();
+
+    // extract host from URL for display name
+    string host = url;
+    size_t pos = url.find("://");
+    if (pos != string::npos) {
+        host = url.substr(pos + 3);
+    }
+
+    providerConfig.id = "ollama-" + to_string(timestamp);


+    // generate unique ID using timestamp
+    auto now = chrono::system_clock::now();
+    auto timestamp = chrono::duration_cast<chrono::seconds>(now.time_since_epoch()).count();
+    providerConfig.id = "openrouter-" + to_string(timestamp);


+    string effectiveKey = apiKey.empty() ? std::string(envKey) : apiKey;
+    try {
+        OpenAiWingman wingman{effectiveKey};
+        vector<string>& models = wingman.listModels();


+        // parse JSon response (OpenAI-compatible format)
+        nlohmann::json httpResponseJSon;
+        try {
+            httpResponseJSon = nlohmann::json::parse(command.httpResponse);
+        } catch (...) {
+            MF_DEBUG(
+                "Error: unable to parse OpenRouter JSon response:" << endl <<
+                "'" << command.httpResponse << "'" << endl
+            );
+
+            command.status = WingmanStatusCode::WINGMAN_STATUS_CODE_ERROR;
+            command.errorMessage = "Error: unable to parse OpenRouter JSon response: '" + command.httpResponse + "'";
+            command.answerMarkdown.clear();
+            command.answerTokens = 0;
+            command.answerLlmModel = llmModel;
+
+            return;
+        }
+
+        MF_DEBUG(
+            "OpenRouterWingman::curlGet() parsed response:" << endl
+            << ">>>"
+            << httpResponseJSon.dump(4)
+            << "<<<"
+            << endl);
+
+        MF_DEBUG("OpenRouterWingman::curlGet() fields:" << endl);
+        if(httpResponseJSon.contains("model")) {
+            httpResponseJSon["model"].get_to(command.answerLlmModel);
+            MF_DEBUG("  model: " << command.answerLlmModel << endl);
+        }
+        if(httpResponseJSon.contains("usage")) {
+            if(httpResponseJSon["usage"].contains("prompt_tokens")) {
+                httpResponseJSon["usage"]["prompt_tokens"].get_to(command.promptTokens);
+                MF_DEBUG("  prompt_tokens: " << command.promptTokens << endl);
+            }
+            if(httpResponseJSon["usage"].contains("completion_tokens")) {
+                httpResponseJSon["usage"]["completion_tokens"].get_to(command.answerTokens);
+                MF_DEBUG("  answer_tokens: " << command.answerTokens << endl);
+            }
+        }
+        if(httpResponseJSon.contains("choices")
+            && httpResponseJSon["choices"].size() > 0
+        ) {
+            auto choice = httpResponseJSon["choices"][0];
+            if(choice.contains("message")
+                && choice["message"].contains("content")
+            ) {
+                choice["message"]["content"].get_to(command.answerMarkdown);
+                m8r::replaceAll(
+                    "\n",
+                    "<br/>",
+                    command.answerMarkdown);
+                MF_DEBUG("  answer (HTML): " << command.answerMarkdown << endl);
+            }
+            if(choice.contains("finish_reason")) {
+                string statusStr{};
+                choice["finish_reason"].get_to(statusStr);
+                if(statusStr == "stop") {
+                    command.status = m8r::WingmanStatusCode::WINGMAN_STATUS_CODE_OK;
+                } else {
+                    command.status = m8r::WingmanStatusCode::WINGMAN_STATUS_CODE_ERROR;
+                    command.errorMessage.assign(
+                        "OpenRouter API HTTP request failed with finish_reason: "
+                        + statusStr);
+                    command.answerMarkdown.clear();
+                    command.answerTokens = 0;
+                    command.answerLlmModel = llmModel;
+                }
+                MF_DEBUG("  status: " << command.status << endl);
+            }
+        } else {
+            command.status = m8r::WingmanStatusCode::WINGMAN_STATUS_CODE_ERROR;
+            command.answerMarkdown.clear();
+            command.answerTokens = 0;
+            command.answerLlmModel = llmModel;
+            if(
+                httpResponseJSon.contains("error")
+                && httpResponseJSon["error"].contains("message")
+            ) {
+                httpResponseJSon["error"]["message"].get_to(command.errorMessage);
+            } else {
+                command.errorMessage.assign(
+                    "No choices in the OpenRouter API HTTP response");
+            }
+        }


+        "OpenRouterWingman::OpenRouterWingman() apiKey: '"
+        << this->apiKey << "'" << endl);


    MF_DEBUG(
        "OpenAiWingman::OpenAiWingman() apiKey: '"
-        << config.getWingmanOpenAiApiKey() << "'" << endl);
+        << this->apiKey << "'" << endl);


+            QMessageBox::information(
+                this,
+                tr("Models Refreshed"),
+                tr("Successfully fetched %1 models from OpenAI API.").arg(models.size()));


…r Qt5+WebEngine macos-latest now maps to macOS 15 on Apple Silicon (ARM64) where Homebrew's qt@5 no longer includes QtWebEngine. Qt5 official builds with WebEngine only exist for x86_64, so switch to: - macos-13 (Intel x86_64) runner - jurplel/install-qt-action@v4 for Qt 5.15.2 with qtwebengine module - keep ccache via brew, remove the now-redundant qt@5 brew install Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

moc processes #ifdef guards with its own limited preprocessor pass. When DO_MF_DEBUG is defined, moc includes doActionMindHack() in the meta-object table unconditionally, but the #ifdef in the .cpp caused the function body to be absent in non-debug builds, yielding: undefined reference to MainWindowPresenter::doActionMindHack() Fix: always declare and define the slot; keep debug logic inside the body under #ifdef DO_MF_DEBUG so both moc and the linker are satisfied regardless of build configuration. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Two fixes: 1. doActionMindHack - restore proper #ifdef DO_MF_DEBUG guards around the slot declaration and implementation so the entire hook is compiled out in production builds (build-rc). All related code in main_menu_view.h/.cpp and main_menu_presenter.cpp was already correctly guarded; the slot itself is now also fully excluded. 2. actionMindSemanticSearch - remove incorrect #ifdef DO_MF_DEBUG guard from its field declaration in main_menu_view.h. This is a real production feature (Wingman semantic search), not a debug hook; the guard caused a build failure when DO_MF_DEBUG is not defined (e.g. build-rc). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 30 out of 31 changed files in this pull request and generated 10 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

    } else {
        // check whether possible
-        if(config.canWingmanOllama()) {
+        if(!config.getLlmProviders().empty()) {


+            if (configDialog.exec() == QDialog::Accepted) {
+                config.addLlmProvider(configDialog.getProviderConfig());
+                refresh();


-
-    return false;
+bool Configuration::isWingman() {
+    return !llmProviders.empty() && !activeLlmProviderId.empty();


+bool Configuration::isWingman() {
+    return !llmProviders.empty() && !activeLlmProviderId.empty();


+    // TODO: actually test the connection by calling OpenAI API
+    // For now, just validate the inputs
+    MF_DEBUG("Configuration::probeOpenAiProvider() validated: " << model << endl);


+    // TODO: actually test the connection by calling ollama API
+    // For now, just validate the inputs
+    MF_DEBUG("Configuration::probeOllamaProvider() validated: " << url << ", " << model << endl);


- */
-bool Configuration::isWingman() {
-    return WingmanLlmProviders::WINGMAN_PROVIDER_NONE==wingmanProvider?false:true;
+    MF_DEBUG("Configuration::probeOpenRouterProvider() validated: " << model << endl);


+        "OpenRouterWingman::OpenRouterWingman() apiKey: '"
+        << this->apiKey << "'" << endl);


…erent colors

…enrouter

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 40 out of 41 changed files in this pull request and generated 14 comments.

    string answerDescriptor{
-        "[provider: " + config.getWingmanLlmProviderAsString(config.getWingmanLlmProvider()) +
+        "[provider: " + config.getLlmProviderById(config.getActiveLlmProviderId())->displayName +
        ", model: " + commandWingmanChat.answerLlmModel +


-
-    return false;
+bool Configuration::isWingman() {
+    return !llmProviders.empty() && !activeLlmProviderId.empty();


+        string effectiveKey = provider->apiKey;
+        if(provider->useEnvVar) {
+            const char* envKey = std::getenv(ENV_VAR_OPENAI_API_KEY);
+            if(envKey) {
+                effectiveKey = string(envKey);
+                MF_DEBUG("  MIND Wingman OpenAI: using env var key" << endl);
+            }
+        }
+        wingman = (Wingman*)new OpenAiWingman{effectiveKey};
+        wingman->setLlmModel(provider->llmModel);


+        string effectiveKey = provider->apiKey;
+        if(provider->useEnvVar) {
+            const char* envKey = std::getenv(ENV_VAR_OPENROUTER_API_KEY);
+            if(envKey) {
+                effectiveKey = string(envKey);
+                MF_DEBUG("  MIND Wingman OpenRouter: using env var key" << endl);
+            }
+        }
+        wingman = (Wingman*)new OpenRouterWingman{effectiveKey};
+        wingman->setLlmModel(provider->llmModel);


+    /**
+     * @brief Get embeddings (not supported by OpenRouter Wingman).
+     */
+    virtual void embeddings(CommandWingmanEmbeddings& command) override {
+        UNUSED_ARG(command);
+        throw std::runtime_error("OpenRouter Wingman does not support embeddings");
+    }


+    string errorMessage;
+    bool success = false;
+
+    if (provider->providerType == WINGMAN_PROVIDER_OPENAI) {
+        success = config.probeOpenAiProvider(
+            provider->apiKey, provider->llmModel, errorMessage);
+    } else if (provider->providerType == WINGMAN_PROVIDER_OLLAMA) {
+        success = config.probeOllamaProvider(
+            provider->url, provider->llmModel, errorMessage);
+    } else if (provider->providerType == WINGMAN_PROVIDER_OPENROUTER) {
+        success = config.probeOpenRouterProvider(
+            provider->apiKey, provider->llmModel, errorMessage);
+    }

-/*
- * Navigator tab
- */
+    if (success) {
+        provider->isValid = true;
+        statusValue->setText(tr("configured"));
+        statusValue->setStyleSheet("QLabel { color: green; }");
+        QMessageBox::information(
+            this,
+            tr("Connection Test"),
+            tr("Provider configuration is valid."));
+    } else {


+            s << CONFIG_SETTING_MIND_WINGMAN_PROVIDER_ITEM
+              << p.id << "|"
+              << p.displayName << "|"
+              << Configuration::getWingmanLlmProviderAsString(p.providerType) << "|"
+              << p.url << "|"
+              << p.apiKey << "|"
+              << p.llmModel << "|"
+              << (p.isValid ? "1" : "0") << "|"
+              << (p.useEnvVar ? "1" : "0") << endl;


+                        // format: id|displayName|typeStr|url|apiKey|llmModel|isValid
+                        std::vector<std::string> parts;


+/**
+ * @brief LLM Provider Configuration
+ * 
+ * Represents configuration for a single Large Language Model provider.
+ * Supports OpenAI and ollama providers with provider-specific fields.
+ */
+struct LlmProviderConfig {
+    std::string id;                    // unique identifier (e.g., "openai-1", "ollama-local")
+    std::string displayName;           // user-friendly name (e.g., "OpenAI GPT-4", "Local Ollama")
+    WingmanLlmProviders providerType;  // WINGMAN_PROVIDER_OPENAI, WINGMAN_PROVIDER_OLLAMA
+    std::string url;                   // for ollama: base URL, for OpenAI: empty
+    std::string apiKey;                // for OpenAI/OpenRouter: stored API key (empty when useEnvVar is true)
+    std::string llmModel;              // model name (e.g., "gpt-4", "llama2")
+    bool isValid;                      // whether configuration was validated/probed
+    bool useEnvVar;                    // if true, resolve API key from environment variable at runtime
+


+                    } else if(line->find(CONFIG_SETTING_MIND_ACTIVE_WINGMAN_PROVIDER) != std::string::npos) {
+                        string id = line->substr(strlen(CONFIG_SETTING_MIND_ACTIVE_WINGMAN_PROVIDER));
+                        c.setActiveLlmProvider(id);
+                    } else if(line->find(CONFIG_SETTING_MIND_WINGMAN_PROVIDER_ITEM) != std::string::npos) {
+                        string v = line->substr(strlen(CONFIG_SETTING_MIND_WINGMAN_PROVIDER_ITEM));
+                        // format: id|displayName|typeStr|url|apiKey|llmModel|isValid
+                        std::vector<std::string> parts;
+                        std::stringstream ss(v);
+                        std::string part;
+                        while(std::getline(ss, part, '|')) {
+                            parts.push_back(part);
+                        }
+                        if(parts.size() >= 7) {
+                            LlmProviderConfig p;
+                            p.id = parts[0];
+                            p.displayName = parts[1];
+                            p.url = parts[3];
+                            p.apiKey = parts[4];
+                            p.llmModel = parts[5];
+                            p.isValid = (parts[6] == "1");
+                            p.useEnvVar = (parts.size() >= 8 && parts[7] == "1");
+                            string typeStr = parts[2];
+                            if(typeStr == Configuration::getWingmanLlmProviderAsString(WingmanLlmProviders::WINGMAN_PROVIDER_OPENAI)) {
+                                p.providerType = WingmanLlmProviders::WINGMAN_PROVIDER_OPENAI;
+                            } else if(typeStr == Configuration::getWingmanLlmProviderAsString(WingmanLlmProviders::WINGMAN_PROVIDER_OLLAMA)) {
+                                p.providerType = WingmanLlmProviders::WINGMAN_PROVIDER_OLLAMA;
+                            } else if(typeStr == Configuration::getWingmanLlmProviderAsString(WingmanLlmProviders::WINGMAN_PROVIDER_OPENROUTER)) {
+                                p.providerType = WingmanLlmProviders::WINGMAN_PROVIDER_OPENROUTER;
+                            } else {
+                                p.providerType = WingmanLlmProviders::WINGMAN_PROVIDER_NONE;
+                            }
+                            c.addLlmProvider(p);
+                        }
+                    }


…a/mindforger into enh-1539/ollama-llm-choice-ii-vibe

feat(app): adding Preferences/Wingman for OpenAI and ollama (vibe coded)

7bede5c

dvorka requested a review from Copilot February 8, 2026 19:20

dvorka self-assigned this Feb 8, 2026

Copilot started reviewing on behalf of dvorka February 8, 2026 19:21 View session

augmentcode Bot reviewed Feb 8, 2026

View reviewed changes

Copilot AI reviewed Feb 8, 2026

View reviewed changes

dvorka added 2 commits May 4, 2026 17:33

chore: 🧹 adding build files to gitignore

3995722

Merge branch 'enh-1539/ollama-llm-choice' into enh-1539/ollama-llm-ch…

5a4a006

…oice-ii-vibe

dvorka changed the title ~~feat(app): adding Preferences/Wingman for OpenAI and ollama (vibe coded)~~ feat(app): ✨ adding Preferences/Wingman for OpenAI and ollama (vibe coded) May 4, 2026

dvorka and others added 5 commits May 4, 2026 17:53

docs(vibe): 📚 removing from Git vibe coded doc

4070f14

refactor(llm): ♻️ refactoring LLM configuration dialog

ae4ac1e

refactor(cfg): ♻️ minor wingman LLM shuffle

6abe363

feat(make): ✨ polishing help target

5a0d090

dvorka requested a review from Copilot May 4, 2026 18:14

Copilot started reviewing on behalf of dvorka May 4, 2026 18:14 View session

Copilot AI reviewed May 4, 2026

View reviewed changes

dvorka added 2 commits May 4, 2026 21:05

build(macos): 🏗️ fixing macos build

6e8f9f0

fix(llm): 🐞 rewriting Ollama code for Win and macOS

5569636

dvorka added the vibe label May 4, 2026

dvorka changed the title ~~feat(app): ✨ adding Preferences/Wingman for OpenAI and ollama (vibe coded)~~ feat(app): ✨ adding Preferences/Wingman for ollama, OpenRouter and OpenAI May 4, 2026

This was linked to issues May 4, 2026

[Feature] note-taking w/ local LLM @ ollama #1539

Open

[Feature] OpenRouter #1559

Open

dvorka and others added 4 commits May 4, 2026 21:16

build(macos): 🏗️ fixing machine & qtinstallation source for build

0bac021

dvorka requested a review from Copilot May 4, 2026 19:40

Copilot started reviewing on behalf of dvorka May 4, 2026 19:41 View session

Copilot AI reviewed May 4, 2026

View reviewed changes

dvorka added 12 commits May 4, 2026 21:47

build(ubuntu): 🏗️ upgrading build to 24.04

6f5fd8d

docs(credits): 📚 adding LLM providers

9f09018

build(ubuntu): 🏗️ GHA ubuntu deps fix

496c076

build(gha): 🏗️ final fix of Ubuntu build

aef814b

feat(mind): ✨ improving thinking to work on big repositories

73d50d8

feat(autolinking): ✨ manual and autolinked links are rendered in diff…

96108dd

…erent colors

feat(tags): ✨ adding colorization for priorities

1a686c9

feat(wingman): ✨ more fixes to Wingman configuration dialog

d496835

fix(config): 🐞 wingman config fixes

bcad72a

fix(wingman): 🐞 fixing keys which had whitespaces around openapi / op…

279d9c6

…enrouter

fix(ui): 🐞 showing LLM provider display name

3370ddc

feat(wingman): ✨ adding privacy warning to config

3306bcd

dvorka requested a review from Copilot May 8, 2026 15:57

Copilot started reviewing on behalf of dvorka May 8, 2026 15:58 View session

test(lib): fix copilot instructions

33b013f

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Copilot AI reviewed May 8, 2026

View reviewed changes

dvorka linked an issue May 8, 2026 that may be closed by this pull request

[Enhancement] Add ability to choose LLM model #1534

Open

dvorka added 2 commits May 8, 2026 21:10

feat(sponsor): ✨ adding sponsor link to UI

db8aee7

Merge branch 'enh-1539/ollama-llm-choice-ii-vibe' of github.com:dvork…

43e1cef

…a/mindforger into enh-1539/ollama-llm-choice-ii-vibe


		#include <QtWidgets>

		#include "../../lib/src/config/configuration.h"

-void Configuration::addLlmProvider(const LlmProviderConfig& provider) {
+void Configuration::addLlmProvider(const LlmProviderConfig& provider) {
+    // Enforce unique provider IDs to avoid ambiguous lookups.
+    if (!provider.id.empty() && getLlmProviderById(provider.id) != nullptr) {
+        MF_DEBUG("Configuration::addLlmProvider() duplicate ID, provider not added: " << provider.id << endl);
+        return;
+    }

-void Configuration::setActiveLlmProvider(const string& id) {
+void Configuration::setActiveLlmProvider(const string& id) {
+    // Only set active provider if the ID is known; keep configuration consistent.
+    if (!id.empty() && getLlmProviderById(id) == nullptr) {
+        MF_DEBUG("Configuration::setActiveLlmProvider() attempted to set unknown provider id: " << id << endl);
+        return;
+    }

		"OpenRouterWingman::OpenRouterWingman() apiKey: '"
		<< this->apiKey << "'" << endl);

		bool Configuration::isWingman() {
		return !llmProviders.empty() && !activeLlmProviderId.empty();

		// format: id\|displayName\|typeStr\|url\|apiKey\|llmModel\|isValid
		std::vector<std::string> parts;

Conversation

dvorka commented Feb 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

augmentcode Bot commented Feb 8, 2026

Uh oh!

augmentcode Bot left a comment

Choose a reason for hiding this comment

Uh oh!

augmentcode Bot Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

augmentcode Bot Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

augmentcode Bot Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

augmentcode Bot Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

augmentcode Bot Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dvorka commented Feb 8, 2026 •

edited

Loading