Content
# spacedrep
[](https://pypi.org/project/spacedrep/)
[](https://pypi.org/project/spacedrep/)
[](LICENSE)
Spaced repetition you can script. A CLI and [MCP server](https://modelcontextprotocol.io/) for flashcards — so an AI agent can generate cards from what you're learning, quiz you in conversation, and manage your deck alongside you.
Import your existing Anki decks to get started. Scheduling uses [FSRS](https://github.com/open-spaced-repetition/py-fsrs) (same algorithm as Anki 23.10+). Everything is JSON in, JSON out — pipe it, script it, or let an agent drive it. [Why build a CLI this way.](https://dev.to/uenyioha/writing-cli-tools-that-ai-agents-actually-want-to-use-39no)
> **You:** Quiz me on my AWS deck.
>
> **Agent:** What does S3 stand for, and what is it used for?
>
> **You:** Simple Storage Service, it's object storage.
>
> **Agent:** Correct. Rated **good** — next review in 4 days. What is the CAP theorem?
>
> **You:** Something about distributed systems... I forget.
>
> **Agent:** CAP: pick 2 of 3 — consistency, availability, partition tolerance. Rated **again** — it'll come back shortly. 12 cards left.
## Install
```bash
pip install spacedrep
```
Optional extras: `spacedrep[mcp]` for the MCP server, `spacedrep[optimizer]` for FSRS parameter optimization (requires torch).
## Features
- **FSRS scheduling** — same algorithm as Anki 23.10+, with parameter optimization from your review history
- **Cloze deletions** — `{{c1::answer}}` syntax, auto-expands into multiple cards
- **Reversed cards** — one note becomes two cards (Q→A and A→Q); edits are template-aware and re-adds dedupe on (question, deck). Basic ↔ reversed collisions are caught with a clear error — delete the existing card first to convert it
- **Anki-native storage** — single SQLite file using Anki's schema for full round-trip compatibility
- **Rich filtering** — by deck, tags, state, date ranges, FSRS properties, and more
- **Agent-friendly** — duplicate detection, leech detection, review preview, bury/unbury, review history
## MCP Server
Add to Claude Desktop (`claude_desktop_config.json`):
```json
{
"mcpServers": {
"spacedrep": {
"command": "uvx",
"args": ["--from", "spacedrep[mcp]", "spacedrep-mcp"],
"env": {
"SPACEDREP_DB": "/path/to/collection.anki21"
}
}
}
}
```
### Agent loop
`get_next_card` → grade → `submit_review`. The server bundles a few affordances for that loop:
- **Loop instructions on connect** — the agent gets the loop, the rating scale, and the rule that it picks the rating (not the user)
- **Inline review history** — `get_next_card` and `get_card` include the last 3 reviews with their `user_answer` and `feedback` text
- **Session pacing** — `get_next_card` returns `due_remaining` for "N cards left" messaging
- **Sibling auto-bury** — `submit_review` buries cloze/reversed siblings for 24h and returns them as `siblings_buried`, so the next card isn't a sibling whose answer was just shown
## CLI Quick Start
```bash
spacedrep db init # create the database
spacedrep deck import ~/Downloads/deck.apkg # or import an existing Anki deck
spacedrep card add "What is CAP theorem?" "Pick 2 of 3: consistency, availability, partition tolerance" --deck AWS
spacedrep card add-cloze "{{c1::Ottawa}} is the capital of {{c2::Canada}}" --deck Geo
spacedrep card add-reversed "Capital of France" "Paris" --deck Geo # creates 2 cards: Q→A and A→Q
spacedrep card next # get the next due card
spacedrep review submit <card_id> good # again | hard | good | easy
```
See `spacedrep --help` and `spacedrep <command> --help` for all options.
## Cloze Deletions
Each `{{c1::answer}}` becomes a separate card. Hints are supported: `{{c1::Ottawa::capital city}}` shows `[capital city]` as the blank.
```bash
spacedrep card update-cloze 42 "{{c1::Ottawa}} is in {{c2::Canada}}, {{c3::North America}}"
```
## Search and Filters
```bash
spacedrep card list --search "Lambda" --deck AWS --state review
spacedrep card list --due-before "2026-12-31" --leeches
spacedrep card bury 42 --hours 4
spacedrep deck export out.apkg
```
Filters work on `card list` and `card next`. See `--help` for the full list.
## Anki Compatibility
spacedrep operates directly on Anki's native SQLite schema. Importing an `.apkg` replaces the working database; exporting writes it back. FSRS scheduling state survives the round-trip — cards reviewed in spacedrep show correct due dates when opened in Anki.
Media files, JavaScript templates, and nested cloze deletions are not supported.
## Data Storage
Your database is a single SQLite file (`collection.anki21` by default). Every card addition and review is committed immediately — if the agent disconnects or the process crashes, no work is lost.
You don't need to export after every session. The `.apkg` import/export tools are for moving data between spacedrep and Anki Desktop, not for saving your progress.
MCP Config
Below is the configuration for this MCP Server. You can copy it directly to Cursor or other MCP clients.
mcp.json
Connection Info
You Might Also Like
everything-claude-code
Complete Claude Code configuration collection - agents, skills, hooks,...
markitdown
Python tool for converting files and office documents to Markdown.
awesome-claude-skills
A curated list of awesome Claude Skills, resources, and tools for...
antigravity-awesome-skills
The Ultimate Collection of 130+ Agentic Skills for Claude...
claude-context-mode
claude-context-mode plugin reduces MCP context bloat, saving up to 99% of tokens.
context-mode
MCP is the protocol for tool access. We're the virtualization layer for context.