Toddyi

toddyi — Toddyi from the terminal

A single binary for people, shell scripts, CI jobs and AI agents: read and change your Toddyi tasks without opening the app.

$ toddyi loginYour code: UCT4-4D8DApprove it at https://toddyi.love/cli/authorize?code=UCT4-4D8DLogged in as [email protected].
$ toddyi create task "Write the quarterly report" --due tomorrow --folder Work --tag q4 --priority highCreated #1038 Write the quarterly report
$ toddyi tasks --due today1040  [~]  2026-10-01  Call the bank1039  [~]  2026-10-01  Buy milk
$ toddyi tasks --due today --ids | toddyi doneDone 2 tasks.

Install

Download the binary for your platform from Releases — one static file, no dependencies — check it against SHA256SUMS, and put it on your PATH.

Or with Go 1.24+:

go install github.com/toddyi-app/toddyi-cli/cmd/toddyi@latest

Building from source:

go build -o ~/.local/bin/toddyi ./cmd/toddyi    # this machinebash build.sh v0.1.0                             # every platform, into dist/

Signing in

toddyi login shows a code and opens the approval page in your browser. You approve it there, signed in as you normally are — second factor included — and the CLI collects a token of its own. The CLI never sees your password.

• The token is stored in ~/.config/toddyi/config.json (mode 0600).
• toddyi logout revokes it on the server, not just locally.
• Every signed-in terminal is listed at /cli/tokens, where any of them can be revoked.
• For CI or an agent that should not write files, set TODDYI_TOKEN (and TODDYI_URL for a server other than toddyi.love) instead of logging in.

Commands

Both word orders work: toddyi create task … and toddyi task create …. Flags may come before or after the arguments. toddyi <command> -h lists a command's flags.

An asset keeps the same ID across all versions. A workspace stores those IDs, version numbers, file paths and SHA-256 checksums in .toddyi-assets.json. sync keeps a local edit if the remote version also changed and reports a conflict instead of overwriting either copy. New files in the workspace are uploaded as private assets; deleting a tracked file creates a recoverable remote deletion, and asset restore ID VERSION can bring it back.

Dates are read in your account's time zone: today, tomorrow, yesterday, +3d, +2w, 2026-10-01, 2026-10-01T09:30, or none. A day without a time is the end of that day for a deadline, so "due today" is not overdue at breakfast.

For scripts and automation

• --json on any command prints JSON on stdout; errors go to stderr as {"error": {"code", "message", "status", "exit_code"}}.
• Exit codes are stable: 0 ok, 1 error, 2 usage or a value the server could not read, 3 not logged in / token revoked, 4 not found, 5 the Free plan's task limit.
• stdin: toddyi create task - makes one task per line; done, reopen, archive, show and delete read ids from stdin when none are given.

# Anything that prints lines becomes tasksgrep -h "TODO" src/**/*.rb | sed 's/.*TODO:\s*//' | toddyi create task - --folder Code --tag todo
# Everything overdue, as JSONtoddyi --json tasks --due overdue | jq -r '.[] | "\(.id)\t\(.title)"'
# A nightly job: archive what was finishedtoddyi tasks --status completed --ids | toddyi archive

For AI agents: toddyi mcp

toddyi mcp is a Model Context Protocol server on stdio. It exposes the same operations as tools — list_tasks, get_task, create_task, update_task, complete_task, reopen_task, archive_task, delete_task, list_folders, create_folder, whoami — and Study tools: list_study_assets, search_study_assets (query plus optional kind, visibility, tag, and created/updated date bounds), get_study_asset, update_study_asset, delete_study_asset, tag_study_asset, restore_study_asset_version, reprocess_study_asset_extraction, analyze_study_asset, list_study_annotations, create_study_annotation, update_study_annotation, delete_study_annotation, and analyze_study_crop. Flashcard tools are list_study_decks, get_study_deck, push_study_deck, clone_study_deck, import_public_study_deck, restore_study_deck_version, and set_study_deck_public. list_study_knowledge_graph, create_study_knowledge_node, update_study_knowledge_node, delete_study_knowledge_node, link_study_knowledge_nodes, and delete_study_knowledge_relation. They use the logged-in account's token and the same owner-scoped server rules. Knowledge nodes can be marked as primary source facts or interpretations and can retain a source fragment ID. Collection and sharing tools include list_study_collections, create_study_collection, update_study_collection, delete_study_collection, create_study_share_link, and revoke_study_share_link. Asset relationships can be browsed with list_study_asset_links, created with create_study_asset_link, and removed with delete_study_asset_link. Fragments and their document hierarchy are available through list_study_fragments and create_study_fragment. MCP publication tools are get_study_publication, create_study_publication, update_study_publication, and delete_study_publication.

Study bundles keep content and provenance while assigning new asset, fragment, annotation, knowledge-node, relation, and publication IDs on import. Existing same-named collections gain the imported asset as a member. Revocable share links are omitted because the token itself grants access and should not be copied into an archive.

Claude Code:

claude mcp add toddyi -- toddyi mcp

Claude Desktop (claude_desktop_config.json), or any MCP client:

{ "mcpServers": { "toddyi": { "command": "toddyi", "args": ["mcp"] } } }

Then ask for things in words: "plan my week from these notes and put it in Toddyi", "what is overdue?", "mark the bank call done". Read-only tools are annotated as such and delete_task as destructive, so a client can ask before using it.

ChatGPT, Claude and Gemini

The remote server offers the everyday tools ? search, fetch, tasks, folders, study materials, decks and the knowledge graph ? and needs nothing installed. toddyi mcp here has the full set, Study editing included.

Releases

Pushing a tag v* runs .github/workflows/release.yml: it tests, builds the CLI for every platform and the ChatGPT bundle, and attaches them and SHA256SUMS to the release. A tag with a hyphen (v0.2.0-rc.1) is published as a pre-release.

Layout

cd cli && go vet ./... && go test -race ./...

The server side is app/controllers/api/cli/ (the API), app/controllers/features/cli/ (the approval page and the token list) and app/models/features/accounts/cli_token.rb.

License

MIT. See LICENSE.