sentra-mcp

# Sentra — mock exams from your PDFs (free, open-source MCP server) Sentra turns lecture PDFs and past exams into realistic mock exams with answer keys. It runs on your machine, costs nothing, and works with any MCP-compatible AI agent (OpenCode, Claude Code, Cursor, VS Code, Gemini CLI). Your agent does the thinking; Sentra does the reading, the analysis prep, and the file saving. Sentra itself makes no model calls. Bilingual site + docs: [`landing/`](landing/) (English / Egyptian Arabic). ## Fastest start 1. Copy this repo link: `https://github.com/mo22zy2/Sentra-MCP` 2. In your AI agent, paste it and say: `download and run this mcp` 3. Restart your agent so the sentra tools appear. 4. Drop lecture PDFs into `lectures/` and past exams into `past_exams/`. 5. Say: `run the sentra pipeline` — the agent follows the prompts in order (`extract_text → knowledge_base → professor_dna → exam_targets → predicted_exam`) and saves each `.md` under `output/`. No forms, no schema — the prompts define the structure. Or say: `make me 2 exams using the sentra mcp`. Not on OpenCode? `uvx sentra-mcp init --agent <claude|cursor|vscode|gemini>` writes the right config for your agent. Every agent uses the same local `lectures/`, `past_exams/`, `output/` folders. ## Prefer the terminal? Install `uv` once (Windows: `winget install -e --id astral-sh.uv`, otherwise: `pip install uv`), then: ```bash uvx sentra-mcp init ``` `init` creates your course folder, connects your agent (backing up its config first), and tells you where to drop PDFs. Then ask your agent: - "Analyze the previous exams and summarize the professor's question patterns." - "Generate a two-hour Operating Systems mock exam with an answer key." You get `output/<name>_exam.md` + `output/<name>_answers.md`, saved locally. ## Your folders ``` <the folder you opened in your agent>/ ├── lectures/ # lecture PDFs (auto-created) ├── past_exams/ # previous exam PDFs (auto-created) └── output/ # gene…