A deep research skill and pipeline for Claude Code and OpenAI Codex that gathers web and scholarly sources into a persistent, searchable markdown vault and writes cited reports.
zaidmukaddam/scira
An open-source, agentic AI research search engine that plans sub-tasks, retrieves live sources, and returns answers with inline citations.
Open-source alternative to:Perplexity
- Stars
- 11,922
- Forks
- 1,500
- Open issues
- 27
- Last commit
- 3 months ago
What is it?
Scira is an AI-powered search and research platform built with Next.js and the Vercel AI SDK. Given a question, it breaks the request into sub-tasks, searches live sources, cross-checks the evidence, and returns an answer with inline citations that link back to each source.
The app offers 17 search modes covering general web search, chat without search, X posts, stocks, academic papers, Reddit, GitHub, crypto, prediction markets, YouTube, and Spotify. It also includes around 28 tools for weather, maps, flights, movies, currency conversion, a sandboxed Python code interpreter, and semantic search over uploaded PDF, CSV, DOCX, and Excel files. Users can choose from a long list of LLMs across xAI, OpenAI, Anthropic, Google, Qwen, Mistral, DeepSeek, and others.
The project is licensed under AGPL-3.0 and can be self-hosted with Docker or run locally with Bun. Running it requires API keys for the model providers and search services it uses.
Key features
- 01Plans multi-step research by breaking questions into sub-tasks and selecting tools.
- 02Returns answers with inline citations that link to the original sources.
- 03Offers 17 search modes, including web, academic, Reddit, GitHub, and YouTube.
- 04Supports many LLM providers, including OpenAI, Anthropic, Google, and DeepSeek.
- 05Includes a sandboxed Python code interpreter and file search over uploaded documents.
- 06Can be self-hosted with Docker or run locally.
Use cases
- Researching a topic across the web, academic papers, and community discussions in one place.
- Checking facts with clickable sources before relying on an AI-generated answer.
- Asking questions about an uploaded PDF or spreadsheet.
- Running a self-hosted AI search engine with your own model API keys.
Quick start
- 1
Create a .env file from .env.example with your API keys
- 2
Run the app with Docker Compose from the project root
docker compose up - 3
Open the app at http://localhost:3000
Things to know
- Requires paid API keys for AI providers such as OpenAI and Anthropic, plus Exa for web search.
- Licensed under AGPL-3.0, which requires source disclosure for modified network-deployed versions.
- Many features depend on numerous third-party API keys, so self-hosting takes setup effort.
In collections
Topics
FAQ
- What is scira?
- An open-source, agentic AI research search engine that plans sub-tasks, retrieves live sources, and returns answers with inline citations. Scira is an AI-powered search and research platform built with Next.js and the Vercel AI SDK. Given a question, it breaks the request into sub-tasks, searches live sources, cross-checks the evidence, and returns an answer with inline citations that link back to each source.
- Is scira free to use? What license does it have?
- scira is open source under the AGPL-3.0 license. Check the license terms before using it in commercial products.
- Is scira still maintained?
- Updated within 6 months. The last commit to scira was on Aug 12, 2026.
- What can I use scira for?
- Researching a topic across the web, academic papers, and community discussions in one place. Checking facts with clickable sources before relying on an AI-generated answer. Asking questions about an uploaded PDF or spreadsheet. Running a self-hosted AI search engine with your own model API keys.
- How do I install scira?
- Create a .env file from .env.example with your API keys Run the app with Docker Compose from the project root: docker compose up Open the app at http://localhost:3000
- Where can I get scira?
- The source code and installation instructions are on GitHub: https://github.com/zaidmukaddam/scira
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Self-hostable AI answering engine that searches the web via SearxNG and cites sources, working with local LLMs through Ollama or cloud providers.
A self-hosted metasearch engine with an AI assistant that can run entirely inside the browser, with no API key required.
Self-hosted AI deep-research web app that plans searches, reads real sources and returns a report with citations you can inspect and follow up on.
AI-powered conversational search engine that answers questions with live web results, supporting OpenAI, Anthropic, Gemini and Vertex AI models.
Local AI search engine that auto-indexes any folder (code, docs, logs, PDFs) and speaks the Elasticsearch API, built in Rust.