image-research-assistant
BrianLusina/image-research-assistant
An Image Research AI powered Assistant
Instalação
npx convez-ai BrianLusina/image-research-assistant1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
Image Research Assistant
This is an image research assistant focused on helping with image-related information retrieval and enrichment using modular services. The repository contains a lightweight application entry point, client/agent scaffolding, and service modules for vision and Wikipedia lookups.
Goals
- Provide a compact, extensible architecture for image research tasks using MCP services.
- Keep services modular so contributors can add new data sources or models.
- Provide simple developer ergonomics for running and testing locally.
Contents
src/app/— main application package and entry pointsrc/app/clients/— client-side agent and MCP client codesrc/app/servers/vision/— vision-related server utilities and serversrc/app/servers/wikipedia/— wikipedia lookup server and helpers
Prerequisites
- macOS/Linux/Windows with a POSIX-like shell (examples use
zsh/bash) - Python 3.10 or newer
- Git
- (Optional)
poetryif you prefer managing the project via Poetry. Thoughuvis the preferred package manager for this project.
Recommended setup
- Create and activate a virtual environment (macOS/Linux):
python3 -m venv .venv
source .venv/bin/activate
If using uv, simply run uv venv after installation to activate the environment.
- Install dependencies. Preferred project methods:
- Use the Makefile helper (recommended):
make install
- Or use
uv(preferred package manager for this project) to install frompyproject.toml:
uv install -r pyproject.toml
If you prefer, you can still perform an editable pip install when iterating on local code:
pip install -e .
Note: There is no requirements.txt in this repository; use pyproject.toml with uv or the Makefile helper.
Running the project
- Run the main application directly (quick):
python src/app/main.py
- Or run as a module after editable install:
make run
Project considerations
- The repository is intentionally modular: add new
servers/orclients/components without touching core logic. - Keep third-party model or API keys out of the repo — use environment variables or a secrets store for credentials.
- Aim for lightweight dependencies to keep the project easy to set up for contributors.
- If adding long-running services, prefer running them behind a process manager (systemd, supervisor) or in containers for production.
Testing
- There are no tests bundled by default. If you add tests, prefer
pytestand place tests undertests/. - Example test run (after adding
pytestto deps):
pytest -q
Contributing
- Fork the repository and create feature branches:
git checkout -b feat/your-feature - Keep commits small and focused, with clear messages.
- Run linters and any tests you add before opening a PR.
- Open a Pull Request against
main, add a descriptive title and summary of changes. - For larger changes, open an issue first to discuss design and approach.
Code style
- Follow idiomatic Python (PEP8). Optionally add
black/ruffor similar tools to the project for consistency. Ruff is used here and can be run usingmake lintand type checking withmake pyrefly.
Contact / Maintainers
- The main repo owner is
BrianLusina(GitHub:BrianLusina). Open issues or PRs for questions, fixes, or feature requests.