The whole backend is one file you download and run

The same idea at a much larger size, built on a real PostgreSQL database rather than SQLite, so it scales further but asks a lot more of you to run yourself.

Also gives you auth, a database, storage and functions behind one API, but it arrives as a set of containers rather than one file.

The same job again, built on PostgreSQL and Hasura, and it hands your data to you as GraphQL instead of REST.

Does the same job of reading, tagging and searching scanned documents, written in Scala and smaller in every way; its last release was March 2025, so check it before relying on it.

The project known as Teedy: document storage with OCR, full-text search, permissions and an approval workflow, aimed more at a small team than at one person.

Not an archive at all, but the tool for everything you do to a PDF before filing it: split, merge, sign, compress, convert.

The original, using the identical store format, so this is the same advice in a different language; pick either and keep the store.

The same encrypted, deduplicated backups, with a desktop window as well as a command, which suits a laptop better than a server.

The long-established answer for backing up Linux machines to a local disk or another machine over SSH, with a different store format that neither of the others can read.

Replaces psql itself with a client that completes table and column names as you type, and it uses pspg as its pager, so the two are partners rather than rivals.

A spreadsheet for the terminal that opens CSV, JSON, SQLite and Excel files and lets you sort, filter and summarise them, so it edits and explores where pspg only displays.

A full SQL editor in the terminal with a schema sidebar and a results pane, which replaces the whole psql plus pager pair rather than improving half of it.

Sends prompts from the command line and logs every prompt and answer to a SQLite file, which makes it the better tool inside scripts where oterm is the better tool for sitting and talking.

Not a competitor but the thing underneath: it downloads and runs open-weight models on your own machine and offers them over a local API.

The same chat idea as a self-hosted web page rather than a terminal, with user accounts and document upload; its licence is a custom one that requires you to keep its branding.

Records a real session you actually type, and stores it as text rather than video, so the file is tiny and the viewer can select the text.

Also records a live session and turns it into an animated GIF or a web player, with the recording edited afterwards in a config file.

The still-picture half by the same authors: it turns a file of code or captured output into a PNG or SVG image, with no motion at all.

Slides from Markdown as well, but built around a live development server with interactive Vue components and a presenter view, so it suits a talk you give rather than a file you send.

The web framework both of the above sit on top of, which you write in HTML instead of Markdown and control completely.

Renders one source file into a deck, an article, a PDF and a website, and runs Python or R code inside the document, which is far more than slides and correspondingly larger to install.

The other end of the job: it holds your subscriber list and sends the campaign, and it expects the HTML this produces.

The same drag-and-drop email editor idea, also as React components, with a hosted version you can use directly rather than embedding.

The answer for people who would rather write the email in code: it lets you build email HTML with Tailwind CSS and handles the awkward conversions for you.

Does the transcription and the speaker labelling together, giving you word-level timings as well, which makes it the one-step answer where pyannote is the specialist.

A general PyTorch toolkit for speech that covers recognition, speaker identification, diarization and enhancement with ready-made recipes, so it is much wider and correspondingly heavier.

NVIDIA's framework for training and running large speech models, including diarization, aimed at people building models rather than using one; it was renamed from NVIDIA/NeMo.

Makes a model return data that matches a schema you define, validating and retrying until it fits, which is the shaping half of the job where DSPy is the tuning half.

A small language in which you declare model functions and their output types, generating typed client code for several languages; it covers structure and types rather than automatic tuning.

Constrains the model as it generates, interleaving your fixed text with patterns the output must match, so the answer cannot come back malformed in the first place.

A deliberately small single-asset backtester with built-in parameter search and interactive charts, easier to start with but without walk-forward analysis as a first-class feature; AGPL-3.0.

Runs the identical strategy code in a backtest and in live trading, with a Rust core, so it is the production answer where PyBroker is the research one.

Evaluates thousands of parameter combinations at once as array operations rather than a loop, which is far faster for sweeps; it carries the same Commons Clause restriction.

Takes a different approach entirely: it sits in front of your site and makes every visitor's browser do a small piece of work first, which costs a scraper running millions of requests far more than it costs a reader.

A much larger Nginx blocklist covering spam referrers, scanners and fake search-engine robots as well as AI ones, so it is broader and noisier.

Google's own parser for robots.txt files, released as a library and a test tool, useful for checking that the rules you wrote say what you think they say.