40df57f969
Problem: - Task Master model configuration wasn't properly checking for API keys in the project's .env file when running through MCP - The isApiKeySet function was only checking session.env and process.env but not inspecting the .env file directly - This caused incorrect API key status reporting in MCP tools even when keys were properly set in .env Solution: - Modified resolveEnvVariable function in utils.js to properly read from .env file at projectRoot - Updated isApiKeySet to correctly pass projectRoot to resolveEnvVariable - Enhanced the key detection logic to have consistent behavior between CLI and MCP contexts - Maintains the correct precedence: session.env → .env file → process.env Testing: - Verified working correctly with both MCP and CLI tools - API keys properly detected in .env file in both contexts - Deleted .cursor/mcp.json to confirm introspection of .env as fallback works
12 lines
2.6 KiB
Plaintext
12 lines
2.6 KiB
Plaintext
# Task ID: 75
|
|
# Title: Integrate Google Search Grounding for Research Role
|
|
# Status: pending
|
|
# Dependencies: None
|
|
# Priority: medium
|
|
# Description: Update the AI service layer to enable Google Search Grounding specifically when a Google model is used in the 'research' role.
|
|
# Details:
|
|
**Goal:** Conditionally enable Google Search Grounding based on the AI role.\n\n**Implementation Plan:**\n\n1. **Modify `ai-services-unified.js`:** Update `generateTextService`, `streamTextService`, and `generateObjectService`.\n2. **Conditional Logic:** Inside these functions, check if `providerName === 'google'` AND `role === 'research'`.\n3. **Construct `providerOptions`:** If the condition is met, create an options object:\n ```javascript\n let providerSpecificOptions = {};\n if (providerName === 'google' && role === 'research') {\n log('info', 'Enabling Google Search Grounding for research role.');\n providerSpecificOptions = {\n google: {\n useSearchGrounding: true,\n // Optional: Add dynamic retrieval for compatible models\n // dynamicRetrievalConfig: { mode: 'MODE_DYNAMIC' } \n }\n };\n }\n ```\n4. **Pass Options to SDK:** Pass `providerSpecificOptions` to the Vercel AI SDK functions (`generateText`, `streamText`, `generateObject`) via the `providerOptions` parameter:\n ```javascript\n const { text, ... } = await generateText({\n // ... other params\n providerOptions: providerSpecificOptions \n });\n ```\n5. **Update `supported-models.json`:** Ensure Google models intended for research (e.g., `gemini-1.5-pro-latest`, `gemini-1.5-flash-latest`) include `'research'` in their `allowed_roles` array.\n\n**Rationale:** This approach maintains the clear separation between 'main' and 'research' roles, ensuring grounding is only activated when explicitly requested via the `--research` flag or when the research model is invoked.
|
|
|
|
# Test Strategy:
|
|
1. Configure a Google model (e.g., gemini-1.5-flash-latest) as the 'research' model in `.taskmasterconfig`.\n2. Run a command with the `--research` flag (e.g., `task-master add-task --prompt='Latest news on AI SDK 4.2' --research`).\n3. Verify logs show 'Enabling Google Search Grounding'.\n4. Check if the task output incorporates recent information.\n5. Configure the same Google model as the 'main' model.\n6. Run a command *without* the `--research` flag.\n7. Verify logs *do not* show grounding being enabled.\n8. Add unit tests to `ai-services-unified.test.js` to verify the conditional logic for adding `providerOptions`. Ensure mocks correctly simulate different roles and providers.
|