agentic research for VS Code
io.github.theaisingularity/agentic-research
Local research agent that verifies its own answers. Runs on Gemma 3 4B + Ollama, $0/query.
client:VS Code
transport:stdio
runtime:pypi
Install agentic research in VS Code
.vscode/mcp.json
{
"inputs": [
{
"type": "promptString",
"id": "openai-base-url",
"description": "OPENAI_BASE_URL",
"password": true
},
{
"type": "promptString",
"id": "openai-api-key",
"description": "OPENAI_API_KEY",
"password": true
},
{
"type": "promptString",
"id": "model-synthesizer",
"description": "MODEL_SYNTHESIZER",
"password": true
},
{
"type": "promptString",
"id": "model-planner",
"description": "MODEL_PLANNER",
"password": true
},
{
"type": "promptString",
"id": "embed-model",
"description": "EMBED_MODEL",
"password": true
},
{
"type": "promptString",
"id": "searxng-url",
"description": "SEARXNG_URL",
"password": true
},
{
"type": "promptString",
"id": "local-corpus-path",
"description": "LOCAL_CORPUS_PATH",
"password": true
},
{
"type": "promptString",
"id": "enable-rerank",
"description": "ENABLE_RERANK",
"password": true
},
{
"type": "promptString",
"id": "enable-fetch",
"description": "ENABLE_FETCH",
"password": true
}
],
"servers": {
"agentic-research": {
"type": "stdio",
"command": "uvx",
"args": [
"agentic-research-engine"
],
"env": {
"OPENAI_BASE_URL": "${input:openai-base-url}",
"OPENAI_API_KEY": "${input:openai-api-key}",
"MODEL_SYNTHESIZER": "${input:model-synthesizer}",
"MODEL_PLANNER": "${input:model-planner}",
"EMBED_MODEL": "${input:embed-model}",
"SEARXNG_URL": "${input:searxng-url}",
"LOCAL_CORPUS_PATH": "${input:local-corpus-path}",
"ENABLE_RERANK": "${input:enable-rerank}",
"ENABLE_FETCH": "${input:enable-fetch}"
}
}
}
}
VS Code prompts for each input the first time the server starts, so no secret is written to the file. VS Code docs