Download the code

Everything this site describes —the service, the panel, the CLI, these very pages— lives in one single public repository:

It is open because a module you cannot obtain is not a module.

This documentation used to hang off antisionista.cl, the first site that uses it, and moved here precisely because the docs belong to the component and not to whoever hosts it. A closed repo left half of that argument unfulfilled: the integration manual is written so that somebody who does not know this code can install it, and it began at a step that person could not execute.

Where it is

Repo	gitlab.com/pineiden/reply2fb
HTTPS	https://gitlab.com/pineiden/reply2fb.git
SSH	git@gitlab.com:pineiden/reply2fb.git
Main branch	main
Latest version	v0.1.0-rc.1 (2026-08-15) — downloads and notes
Licence	AGPL-3.0-or-later — see below

The repo is called reply2fb and the module reply2social

That is not your mistake. The project was born mirroring Instagram videos to a Facebook Page —hence reply2fb— and grew into a multi-network system: Mastodon, Telegram, RSS and your own site. The repo name stayed for compatibility with the clones and remotes that already exist; the module, its npm packages (@reply2social/core, @reply2social/svelte) and this documentation are reply2social. You will also see reply2fb in the binary, in the database name and in the environment variables.

Downloading it

Clone it naming the destination directory. This is not cosmetic: the compose.yml in the integration manual points at ./reply2social/backend, and a git clone without a destination leaves you a reply2fb directory that does not match.

git clone https://gitlab.com/pineiden/reply2fb.git reply2social
cd reply2social

Verify

git log --oneline -1        # must show a commit
ls backend/Cargo.toml ui/core/package.json LICENSE

The three files must exist. If LICENSE is missing you are on an old clone: bring the changes in with git pull.

No GitLab account, no git

If you only want to read it or build it once, you need neither an account nor a git client:

curl -L -o reply2social.tar.gz \
  https://gitlab.com/pineiden/reply2fb/-/archive/main/reply2fb-main.tar.gz
tar xzf reply2social.tar.gz
mv reply2fb-main reply2social

That gets you main, which moves. If you are going to install it for real, download a tagged version instead of main — that is the next section.

Tagged versions

If you are installing this into something that matters to someone, do not install main: install a version. Each one has its entry in the CHANGELOG, which says what it ships and —more importantly— what it did not verify.

Version	Date	What it is	Download
v0.1.0-rc.1	2026-08-15	First tagged version. Release candidate: Instagram never ran against Meta. Notes	.tar.gz · .zip

The always-current list: gitlab.com/pineiden/reply2fb/-/tags

With git:

# See which versions exist
git tag --list --sort=-v:refname

# Stay on one
git checkout v0.1.0-rc.1

Without git:

curl -L -o reply2social-v0.1.0-rc.1.tar.gz \
  https://gitlab.com/pineiden/reply2fb/-/archive/v0.1.0-rc.1/reply2fb-v0.1.0-rc.1.tar.gz
tar xzf reply2social-v0.1.0-rc.1.tar.gz
mv reply2fb-v0.1.0-rc.1 reply2social

What the -rc means, here

It is neither modesty nor ritual. It means there is a part of the system that could not be verified against reality: Instagram, the project’s original source, has never run against Meta’s API — what is missing is the one thing you cannot program, the tokens. That entire path is tested against a fake Graph running inside the tests, and a fake server cannot contradict what a real one would say.

The CHANGELOG lists the seven limits of this class, one by one. Read them before deciding whether this goes into your site. And while the version is 0.x, any release may break backwards compatibility.

main is not a version. It moves, and it can break backwards compatibility without warning. If you are going to install from main anyway, write down the sha shown at /-/commits/main the day you downloaded it, and pin it with git checkout <sha>. Without a version marker, a green build may be running the previous code and nobody finds out.

What is inside

reply2social/
├── backend/          the service: Rust + Postgres, and the `reply2fb` CLI
│   ├── src/          poller, ingest, archive, emission, networks
│   ├── migrations/   the schema, in order
│   └── Dockerfile    what the manual's compose builds
├── ui/
│   ├── core/         @reply2social/core — HTTP client + rules, framework-free
│   ├── svelte/       @reply2social/svelte — the eight screens
│   └── empaquetar.mjs  generates the .tgz files of step 6.1
├── docs/             the design, the decisions and this site (docs/sitio/)
├── compose.yml       the reference deployment
├── .env.example      every variable, commented
├── CHANGELOG.md      what each version ships, and what it did NOT verify
└── LICENSE           AGPL-3.0

You do not need all of it. Depending on what you are doing:

If you want to…	You only need	Continue with
Run the service and talk to it over HTTP from your own front end	backend/, compose.yml, .env.example	Integrate, steps 0 to 5
Also mount the eight screens	add ui/	Integrate, step 6
Understand how it is built before installing anything	docs/ and this site	The modules
Build these pages locally	docs/sitio/	below

Build this documentation locally

cd docs/sitio
npm ci
npm run build      # builds and RUNS the verification
npm run dev        # local server

npm run build does not end at astro build: afterwards it runs scripts/verificar.mjs, which fails if any page came out empty or if a diagram did not render. It exists because two diagrams with a syntax error were published as blank pages —with title, menu and footer, and nothing in between— and the build said “Complete”.

The licence: AGPL-3.0

reply2social is published under the GNU Affero General Public License, version 3 or later. The full text is in LICENSE, and it is also declared by backend/Cargo.toml and by both package.json files in ui/.

In practical terms:

You may	On the condition of
Use it, for anything, commercially included	—
Study it and modify it	—
Install it on your site and run it	—
Redistribute it, modified or not	publishing the source code, under the same licence
Offer it as a network service, distributing nothing	publishing your modifications anyway to whoever uses it

Why AGPL and not MIT, nor even GPL.

That last row is the whole difference. Under the GPL, someone can take this code, modify it, and offer it as a hosted service without publishing a single line: they never distribute the binary, so the obligation never triggers. The AGPL closes that door — it is clause 13, “Remote Network Interaction”.

This is a service: it is exactly the case the GPL leaves open. And what the module does —archiving other people’s material and republishing it with permission— depends on its guards being auditable. A closed derivative running on somebody else’s server would be a system deciding about third-party material and about consents, that nobody can review.

What the licence requires of you if you deploy it

If you run a modified version and give other people access to it over a network, you must offer them the source of that version. The usual way to comply is to publish your fork and link to it from the interface. If you run it unmodified, this repository is enough.

Reporting something, or proposing a change

• Issues: gitlab.com/pineiden/reply2fb/-/work_items — yes, work_items and not issues: GitLab migrated this project to work items and the old path returns 404.
• Merge requests: gitlab.com/pineiden/reply2fb/-/merge_requests. They come from a fork of the repo and are accepted under the same licence.

If you are reporting an integration failure, the most useful thing you can attach is which verification from the manual failed and what it returned. The manual is written so that every step has one: a bare “it doesn’t work” forces us to guess which of the eight silent failure modes it was.

Do not open a public issue containing a token, an encryption key or a .env. If you find a security problem, write privately to whoever maintains the repo instead of opening an issue.

And now

• Step by step in your site — the integration manual, with a verification for every step. Start here if you already downloaded the code.
• A post’s journey — what the system does, before installing it.
• The modules — how what you just cloned is split up.