SARAL AI: Simplified And Automated Research Amplification and Learning

SARAL AI is a full-stack application that transforms research papers (PDF, arXiv, BioRxiv, patents) into AI-generated video presentations, podcasts, posters, reels, slides, and business briefs. It leverages a Go gateway orchestrator, seven workers across five services, and a Next.js 16 frontend with Firebase authentication to deliver a seamless pipeline from document upload to downloadable media.

Developed at Precog Labs, IIIT Hyderabad, this national initiative by the Anusandhan National Research Foundation (ANRF) aims to democratize knowledge and improve how research is consumed. Recognized by Indian Ministry of Science and Technology, featured on Nature article, and Analytics India Magazine (AIM).

Repository Layout

saral/
├── backend/
│   ├── gateway/                 Go REST API + orchestrator (port 8080)
│   │   ├── internal/            auth, pipeline, db, sse, gemini, sarvam, …
│   │   ├── .env                 ← you create (gateway-only config)
│   │   └── firebase-service-account.json   ← you download (secret)
│   ├── services/
│   │   ├── pdf-parser/          Python worker — PDF text/image extraction
│   │   ├── beamer/              Python worker — LaTeX slides + posters (2 processes)
│   │   ├── ffmpeg-job/          Python worker — video stitching
│   │   ├── script-gen/          Go worker — Gemini script generation
│   │   ├── audio-gen/           Go worker — Sarvam/Bhashini TTS
│   │   └── shared/              Python shared lib (saral_shared)
│   ├── migrations/              Plain SQL, applied with psql
│   ├── docker-compose.yml       Postgres + Redis + fake-GCS
│   ├── Procfile.dev             overmind process definitions
│   ├── .env.shared              ← you create (shared by ALL workers)
│   ├── ROUTES.md                Full API reference
│   └── ARCHITECTURE.md          System design & data flow
└── frontend/
    ├── app/                     Next.js 16 App Router pages
    ├── components/              ui (shadcn), dashboard, modals, landing, auth
    ├── lib/                     api client, zustand stores, firebase, types
    └── .env.local               ← you create (Firebase web config)

Architecture & User Flow:

The Next.js frontend (port 3000) talks to the Go gateway (port 8080).
The gateway authenticates via Firebase (optional locally), stores state in Postgres, uploads artifacts to GCS (fake-GCS locally), and enqueues jobs on Redis Streams.
Seven workers consume those streams and report back via webhooks; the gateway pushes progress to the browser over SSE.

flowchart TD
    User([User]) -->|Uploads Paper PDF| UI[Next.js Frontend]
    UI -->|API POST /upload| Gateway[Go Gateway API]
    Gateway -->|Save File| Storage[(Storage: GCS / fake-GCS)]
    Gateway -->|Enqueue Task| Redis[Redis Streams]
    
    subgraph Workers
        PDF[pdf-parser]
        Script[script-gen]
        Audio[audio-gen]
        Beamer[beamer_compile]
        FFmpeg[ffmpeg_stitch]
    end
    
    Redis -->|1. Parse| PDF
    PDF -->|Webhook| Gateway
    
    Gateway -->|2. Generate| Script
    Script -->|Webhook| Gateway
    
    Gateway -.->|3. Pause for human review| UI
    UI -->|Confirm Script| Gateway
    
    Gateway -->|4. Parallel Processing| Audio
    Gateway -->|4. Parallel Processing| Beamer
    
    Audio -->|Webhook| Gateway
    Beamer -->|Webhook| Gateway
    
    Gateway -->|5. Stitch Media| FFmpeg
    FFmpeg -->|Webhook| Gateway
    
    Gateway -->|SSE Progress| UI
    UI -->|Download video.mp4| User

For the deeper architecture guide, see backend/ARCHITECTURE.md.

1. Prerequisites

Step 1: Install Package Managers & Core Setup

macOS:

Ensure you have Homebrew installed.

Windows (WSL2):

You must use WSL2 (Windows Subsystem for Linux). Do not try to run overmind or poppler on native Windows.

Open PowerShell as Administrator and run wsl --install.
Open your new Ubuntu terminal and install Homebrew for Linux:

# 1. Install Homebrew for Linux
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

# 2. Add Homebrew to your PATH
(echo; echo 'eval "$(/home/linuxbrew/.linuxbrew/bin/brew shellenv)"') >> ~/.bashrc
eval "$(/home/linuxbrew/.linuxbrew/bin/brew shellenv)"

# 3. Install GCC (required by Homebrew on Linux)
sudo apt-get update && sudo apt-get install -y build-essential

Linux (Ubuntu/Debian): Follow the same Homebrew installation block as WSL2 above, OR use your native package manager for the dependencies below.

Step 2: Install Core Dependencies

Once Homebrew is available (macOS, Linux, or WSL2):

# Core toolchain
brew install go python@3.11 uv overmind redis postgresql node

# Worker system dependencies
brew install ffmpeg poppler            # ffmpeg-job + beamer (pdf2image)

Step 3: Install LaTeX (for Beamer)

macOS:

# Option A (full distribution):
brew install --cask mactex-no-gui
sudo tlmgr update --self
sudo tlmgr install beamer latexmk

# Option B (smaller install):
brew install --cask basictex
export PATH="/Library/TeX/texbin:$PATH"
sudo tlmgr update --self
sudo tlmgr install beamer collection-latexrecommended collection-fontsrecommended xetex latexmk

Windows (WSL2) & Linux:

sudo apt-get install -y texlive-xetex texlive-fonts-recommended texlive-latex-recommended

Step 4: Docker & Verify

Install Docker Desktop (enable WSL2 integration on Windows).

Verify everything is ready:

go version            # ≥ 1.25
python3.11 --version  # 3.11.x
docker compose version
overmind --version
xelatex --version     # after terminal restart
ffmpeg -version
node --version        # ≥ 20

API keys you need

Key	Where to get it	Used by
Gemini API key	https://aistudio.google.com/app/apikey	script-gen, audio-gen, gateway
OpenRouter API key	https://openrouter.ai/keys	script-gen (when `LLM_PROVIDER=openrouter`)
Sarvam API key	https://dashboard.sarvam.ai	audio-gen (TTS)
LinkedIn API keys	https://www.linkedin.com/developers/apps — create an app, request 'Share on LinkedIn' and 'Sign In with LinkedIn'	gateway (social publishing)
YouTube API keys	https://console.cloud.google.com/apis/credentials — create OAuth 2.0 Client IDs, enable YouTube Data API v3	gateway (social publishing)

2. Backend Setup

All commands from saral/backend/.

2.1 Start infrastructure

docker compose up -d
docker compose ps     # wait for postgres + redis → "healthy"

Docker Permission Denied Error (WSL/Linux)? If you get permission denied while trying to connect to the docker API at unix:///var/run/docker.sock, you can either run the commands with sudo (e.g. sudo docker compose up -d), or permanently fix it by adding your user to the docker group:
sudo usermod -aG docker $USER
newgrp docker

This starts (host ports):

Container	Host port	Credentials / notes
postgres:15-alpine	5433	user `saral_app`, password `localpassword`, db `saral`. `init.sql` auto-applies on first boot.
redis:7-alpine	6380	job streams
fake-gcs-server	4443	bucket `saral-artifacts-local` auto-created by the `fake-gcs-init` container

The DB password is localpassword (defined in docker-compose.yml). The old backend/SETUP.md says saral_pass — that is wrong.

Health checks:

psql "postgresql://saral_app:localpassword@localhost:5433/saral" -c "SELECT 1;"
redis-cli -p 6380 ping                                  # PONG
curl -s http://localhost:4443/storage/v1/b | head -5    # JSON

2.2 Apply numbered migrations

init.sql is applied automatically by the container. The numbered ones are not — apply manually, in order:

for f in migrations/0*.sql; do
  echo "== $f"
  psql "postgresql://saral_app:localpassword@localhost:5433/saral" -f "$f"
done

Verify: psql "postgresql://saral_app:localpassword@localhost:5433/saral" -c "\dt" → should list users, papers, pipeline_runs, pipeline_steps, artifacts, plus podcast/reel/slides/social/patent tables.

2.3 Create `backend/.env.shared`

Loaded by all six workers (Python workers via load_dotenv(../../.env.shared), Go workers via godotenv.Load("../../.env.shared")). The gateway does not read this file.

# ── Infrastructure (matches docker-compose.yml) ─────────────────
ENV=local
DATABASE_URL=postgresql://saral_app:localpassword@localhost:5433/saral
REDIS_URL=redis://localhost:6380
STORAGE_BUCKET=saral-artifacts-local
STORAGE_EMULATOR_HOST=http://localhost:4443

# ── Worker → gateway callbacks ──────────────────────────────────
GATEWAY_WEBHOOK_URL=http://localhost:8080
FRONTEND_BASE_URL=http://localhost:3000

# ── External APIs (REQUIRED — pipeline dies without these) ─────
GEMINI_API_KEY=<your-gemini-key>          # script_gen step fails if missing
SARVAM_API_KEY=<your-sarvam-key>          # audio_gen step fails if missing
GCP_PROJECT_ID=<your-gcp-project-id>

# ── Translation (sarvam is the local-dev default) ──────────────
TRANSLATION_PROVIDER=sarvam

# ── Social publishing (optional — leave blank locally) ─────────
LINKEDIN_CLIENT_ID=
LINKEDIN_CLIENT_SECRET=
LINKEDIN_REDIRECT_URI=

2.3.1 Choosing an LLM provider (script-gen)

The script-gen worker is LLM-agnostic. Pick a backend via LLM_PROVIDER in backend/.env.shared:

Mode	When to use	Required env vars
`vertex` (default)	Existing GCP deployments; uses Vertex AI Gemini	`GCP_PROJECT_ID`, ADC
`gemini_api`	Quick Gemini access via direct API key	`GEMINI_API_KEY`
`openrouter`	Any model on OpenRouter (free tier, GPT, Claude, Llama, etc.)	`OPENROUTER_API_KEY` (+ optional model overrides)

Example openrouter block (uncomment in .env.shared and fill in your own key):

LLM_PROVIDER=openrouter
OPENROUTER_API_KEY=sk-or-v1-...
OPENROUTER_FLASH_MODEL=openrouter/free
OPENROUTER_PRO_MODEL=openai/gpt-oss-120b:free
OPENROUTER_SITE_URL=http://localhost:3000
OPENROUTER_SITE_NAME="Saral Dev"

Any OpenRouter model slug works for OPENROUTER_FLASH_MODEL and OPENROUTER_PRO_MODEL. The HTTP-Referer (OPENROUTER_SITE_URL) and X-Title (OPENROUTER_SITE_NAME) headers are recommended by OpenRouter for rankings and attribution.

2.4 Create `backend/gateway/.env`

Loaded only by the gateway (its working directory is gateway/, so relative paths resolve from there):

ENV=local
PORT=8080
DATABASE_URL=postgresql://saral_app:localpassword@localhost:5433/saral
REDIS_URL=redis://localhost:6380
STORAGE_BUCKET=saral-artifacts-local
STORAGE_EMULATOR_HOST=http://localhost:4443
FRONTEND_BASE_URL=http://localhost:3000

# ── GCP (prod only — blank locally) ────────────────────────────
GCP_PROJECT_ID=
GCP_REGION=asia-south1

# ── Per-user API key encryption (users table stores encrypted
#    gemini/sarvam keys) — generate: openssl rand -base64 32 ────
KEYS_ENCRYPTION_KEY=<random-32-byte-base64>

# ── Optional social integrations — blank locally ───────────────
SARVAM_API_KEY=<your-sarvam-key>
YOUTUBE_CLIENT_ID=
YOUTUBE_CLIENT_SECRET=
YOUTUBE_REDIRECT_URI=

2.5 Place required secret/data files

File	What	How
`gateway/internal/auth/common_passwords.txt`	SecLists 10k list, embedded at compile time	If missing: `curl -s https://raw.githubusercontent.com/danielmiessler/SecLists/master/Passwords/Common-Credentials/10k-most-common.txt -o gateway/internal/auth/common_passwords.txt`
`services/audio-gen/models.json`	Bhashini model registry	A stub file containing just `[]` is fine when `TRANSLATION_PROVIDER=sarvam`. Real Bhashini credentials only needed for the 10 extended Indic languages.

2.6 Install dependencies

# Go gateway
cd gateway && go mod tidy && cd ..

# Python workers — ONLY these four are Python (script-gen & audio-gen are Go)
# UV_LINK_MODE=copy prevents permission errors on WSL/Windows cross-filesystem mounts
export UV_LINK_MODE=copy
for svc in pdf-parser beamer ffmpeg-job shared; do
  cd services/$svc
  uv sync --python 3.11
  cd ../..
done

The Go workers (script-gen, audio-gen) need no install step — go run resolves modules on first start.

2.7 Start everything

overmind start -f Procfile.dev

** WSL Error: bind: operation not supported?** Overmind uses Unix sockets, which Windows file systems (/mnt/c/) do not natively support. You must tell Overmind to place its socket file on the Linux side (like /tmp/):
OVERMIND_SOCKET=/tmp/overmind.sock overmind start -f Procfile.dev

Seven processes start: gateway, pdf-parser, latex-worker, poster-worker, ffmpeg-worker, script-gen, audio-gen.

Overmind kills all processes if any one crashes. When reading crash logs, find the first Exited with code 1 — everything after it showing KeyboardInterrupt / signal: interrupt is collateral, not a real error.

2.8 Verify

curl http://localhost:8080/health
# → {"success":true,"data":{"status":"ok"},...}

# Upload a test paper using the local auth bypass (no Firebase needed):
curl -X POST http://localhost:8080/api/papertovideo/upload \
  -H "X-User-ID: testuser1" \
  -F "pdf=@/path/to/paper.pdf"

# Watch it process live:
curl -N http://localhost:8080/api/papertovideo/<run_id>/stream -H "X-User-ID: testuser1"

The pipeline pauses after script_gen for human review — confirm to continue:

curl -X POST http://localhost:8080/api/papertovideo/<run_id>/script/confirm \
  -H "X-User-ID: testuser1" -H "Content-Type: application/json" -d '{}'

Full endpoint walkthrough: backend/ROUTES.md.

3. Frontend Setup

All commands from saral/frontend/.

3.1 Create `frontend/.env.local`

Values from Firebase Console → Project Settings → General → Your apps → Web app (create a web app if none exists):

NEXT_PUBLIC_FIREBASE_API_KEY=<apiKey>
NEXT_PUBLIC_FIREBASE_AUTH_DOMAIN=<project>.firebaseapp.com
NEXT_PUBLIC_FIREBASE_PROJECT_ID=<projectId>
NEXT_PUBLIC_FIREBASE_STORAGE_BUCKET=<project>.appspot.com
NEXT_PUBLIC_FIREBASE_MESSAGING_SENDER_ID=<senderId>
NEXT_PUBLIC_FIREBASE_APP_ID=<appId>

# Gateway URL (this is also the default if unset)
NEXT_PUBLIC_GATEWAY=http://localhost:8080

NEXT_PUBLIC_* vars are inlined at build time — restart the dev server after changing them.

3.2 Install & run

bun install     # postinstall copies pdfjs worker → public/pdfjs/
bun run dev     # → http://localhost:3000

3.3 Frontend stack cheat-sheet

Layer	Choice
Framework	Next.js 16 App Router, React 19, TypeScript
Styling	Tailwind v4, tokens in `app/globals.css` `@theme inline`
Components	shadcn (`components/ui/`) — add via `npx shadcn@latest add <name>`
State	zustand: `lib/auth-store.ts`, `lib/paper-store.ts`, `lib/artifact-store.ts`
API client	`lib/api.ts` (all gateway calls), types in `lib/types.ts`
Auth	Firebase client SDK (`lib/firebase.ts`); guard in `app/dashboard/layout.tsx`
Theming	next-themes via `context/ThemeContext.tsx` (light default, class-based dark)

4. Daily Workflow

# Terminal 1 — infra (once; survives reboots until `docker compose down`)
cd saral/backend && docker compose up -d

# Terminal 2 — backend services
cd saral/backend && overmind start -f Procfile.dev

# Terminal 3 — frontend
cd saral/frontend && bun run dev

Task	Command
Logs for one service	`overmind connect gateway` (Ctrl+B, D to detach)
Restart one service	`overmind restart latex-worker`
Stop backend services	`overmind kill`
Stop infra (keep data)	`docker compose down`
Stop infra (wipe data)	`docker compose down -v` (numbered migrations must be re-applied after)

5. Ports Reference

Service	Port	What
Frontend (Next.js)	3000	Web UI
Gateway (Go/Gin)	8080	REST API + SSE
PostgreSQL	5433	State (container 5432)
Redis	6380	Job streams (container 6379)
fake-GCS	4443	Artifact storage emulator

6. Troubleshooting

Backend won't start

Error	Fix
`pattern common_passwords.txt: no matching files found`	Download SecLists file → `gateway/internal/auth/common_passwords.txt` (see §2.5)
`bhashini registry: read models.json: no such file`	Create `services/audio-gen/models.json` with `[]` (see §2.5)
`Firebase init failed: cannot read credentials file`	Service-account JSON missing at path in `FIREBASE_CREDENTIALS_FILE` (relative to `gateway/`)
`Firebase init failed: no private key data found`	The JSON is a placeholder — download the real key from Firebase Console
`dial tcp [::1]:6380: connect: connection refused`	Infra not running or ports unbound: `docker compose ps` must show `0.0.0.0:6380->6379`. If not: `docker compose down && docker compose up -d` from `backend/`
Python `ModuleNotFoundError`	Recreate that worker's venv (§2.6). Note: only 4 services are Python — running `uv sync` in `script-gen`/`audio-gen` errors with "No pyproject.toml" (harmless)
`xelatex failed` / blank slides	`brew install --cask basictex`, restart terminal
ffmpeg-worker fails to stitch	`brew install ffmpeg`
pdf2image / poppler errors	`brew install poppler`

Pipeline runs but a stage fails

Check error_message: curl -s localhost:8080/api/papertovideo/<run_id>/status -H "X-User-ID: testuser1"

Stage	Usual cause
`script_gen`	`GEMINI_API_KEY` missing/invalid in `backend/.env.shared`, or (when `LLM_PROVIDER=openrouter`) `OPENROUTER_API_KEY` missing/invalid (`OPENROUTER_API_KEY is required when LLM_PROVIDER=openrouter`)
`audio_gen`	`SARVAM_API_KEY` missing/invalid in `backend/.env.shared`
`beamer_compile`	xelatex missing, or font issues (check `overmind connect latex-worker`)
`ffmpeg_stitch`	ffmpeg missing
Pipeline "stuck" after script_gen	Not stuck — it's the intentional human-review pause. `POST /script/confirm`.

Frontend issues

Symptom	Fix
`auth/invalid-api-key` in browser console	`.env.local` Firebase values missing/wrong; restart dev server after editing
API calls hit wrong host	Set `NEXT_PUBLIC_GATEWAY=http://localhost:8080` and restart
PDF previews broken	`public/pdfjs/pdf.worker.min.mjs` missing — re-run `npm install` (postinstall copies it)
Login works but dashboard kicks you out	Backend gateway not running, or `FIREBASE_PROJECT_ID` mismatch between gateway `.env` and frontend `.env.local`

Redis / queue debugging

redis-cli -p 6380 XLEN saral:jobs:pdf            # queue depth
redis-cli -p 6380 XINFO GROUPS saral:jobs:pdf    # consumer lag
redis-cli -p 6380 XRANGE saral:dlq - + COUNT 10  # dead letters

7. Environment Variable Reference

`backend/.env.shared` (all workers)

Variable	Required	Purpose
`ENV`	yes	`local` / `production`
`DATABASE_URL`	yes	Postgres DSN (workers write step state)
`REDIS_URL`	yes	Job streams
`STORAGE_BUCKET`	yes	`saral-artifacts-local` locally
`STORAGE_EMULATOR_HOST`	local only	`http://localhost:4443` — unset in prod
`GATEWAY_WEBHOOK_URL`	yes	Where workers POST completion webhooks
`LLM_PROVIDER`	no	LLM backend: `vertex` (default), `gemini_api`, or `openrouter`
`GEMINI_API_KEY`	depends	Required when `LLM_PROVIDER=gemini_api`; also used by audio-gen
`OPENROUTER_API_KEY`	when `LLM_PROVIDER=openrouter`	OpenRouter bearer token
`OPENROUTER_FLASH_MODEL`	no	OpenRouter model slug for Flash tier; default `google/gemini-2.5-flash`
`OPENROUTER_PRO_MODEL`	no	OpenRouter model slug for Pro tier; default `google/gemini-2.5-pro`
`OPENROUTER_SITE_URL`	no	`HTTP-Referer` header sent to OpenRouter
`OPENROUTER_SITE_NAME`	no	`X-Title` header sent to OpenRouter
`SARVAM_API_KEY`	yes	audio-gen TTS
`GCP_PROJECT_ID`	yes	Gemini/Vertex project
`TRANSLATION_PROVIDER`	no	`sarvam` (default) / `bhashini`
`MODELS_JSON_PATH`	no	Override path to audio-gen models.json
`FRONTEND_BASE_URL`	no	Used in links/redirects
`LINKEDIN_*`	no	Social publishing (blank locally)

`backend/gateway/.env` (gateway only)

Variable	Required	Purpose
`ENV`, `PORT`	yes	`local`, `8080`
`DATABASE_URL`, `REDIS_URL`	yes	Same values as .env.shared
`STORAGE_BUCKET`, `STORAGE_EMULATOR_HOST`	yes	Artifact storage
`FIREBASE_PROJECT_ID`	yes	Token verification
`FIREBASE_CREDENTIALS_FILE`	yes	Path to service-account JSON (relative to `gateway/`)
`FIREBASE_WEB_API_KEY`	for real auth	Email/password + Google OAuth REST flows
`KEYS_ENCRYPTION_KEY`	yes	Encrypts per-user API keys stored in `users` table
`GEMINI_MODEL`	no	Override default Gemini model
`SARAL_ADMIN_UIDS`	no	Comma-separated admin Firebase UIDs
`GCP_PROJECT_ID`, `GCP_REGION`	prod only	Cloud Run deployment
`YOUTUBE_`, `LINKEDIN_`, `DASHBOARD_WEBHOOK_*`	no	Integrations (blank locally)

`frontend/.env.local`

Variable	Required	Purpose
`NEXT_PUBLIC_FIREBASE_API_KEY`	yes	Firebase web SDK
`NEXT_PUBLIC_FIREBASE_AUTH_DOMAIN`	yes	〃
`NEXT_PUBLIC_FIREBASE_PROJECT_ID`	yes	〃 (must match gateway's `FIREBASE_PROJECT_ID`)
`NEXT_PUBLIC_FIREBASE_STORAGE_BUCKET`	yes	〃
`NEXT_PUBLIC_FIREBASE_MESSAGING_SENDER_ID`	yes	〃
`NEXT_PUBLIC_FIREBASE_APP_ID`	yes	〃
`NEXT_PUBLIC_GATEWAY`	no	Gateway URL; defaults to `http://localhost:8080` (`NEXT_PUBLIC_API_URL` also accepted)

8. Further Reading

backend/ARCHITECTURE.md — system diagrams, data flow, scaling
backend/AUTH_ENDPOINTS.md — auth contract details
backend/Saral Overhaul API Collection/ — importable API collection

9. Notes for Contributors

Never commit secrets. Keep .env*, firebase-service-account.json, and *firebase*.json out of version control — they are already in .gitignore.
Document new environment variables in both .env.shared / .env.local and in this README immediately.
Update this guide if you add a new local dependency or setup step.
Run linters before pushing: go vet ./..., ruff check ., bun run lint, npx tsc --noEmit.
For full contribution guidelines, code of conduct, and PR checklist, see CONTRIBUTING.md.

Name		Name	Last commit message	Last commit date
Latest commit History 39 Commits
.github		.github
backend		backend
frontend		frontend
.gitignore		.gitignore
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
README.md		README.md

Uh oh!

Folders and files

Latest commit

History

Repository files navigation

SARAL AI: Simplified And Automated Research Amplification and Learning

Repository Layout

Table of Contents

1. Prerequisites

Step 1: Install Package Managers & Core Setup

Step 2: Install Core Dependencies

Step 3: Install LaTeX (for Beamer)

Step 4: Docker & Verify

API keys you need

2. Backend Setup

2.1 Start infrastructure

2.2 Apply numbered migrations

2.3 Create backend/.env.shared

2.3.1 Choosing an LLM provider (script-gen)

2.4 Create backend/gateway/.env

2.5 Place required secret/data files

2.6 Install dependencies

2.7 Start everything

2.8 Verify

3. Frontend Setup

3.1 Create frontend/.env.local

3.2 Install & run

3.3 Frontend stack cheat-sheet

4. Daily Workflow

5. Ports Reference

6. Troubleshooting

Backend won't start

Pipeline runs but a stage fails

Frontend issues

Redis / queue debugging

7. Environment Variable Reference

backend/.env.shared (all workers)

backend/gateway/.env (gateway only)

frontend/.env.local

8. Further Reading

9. Notes for Contributors

About

Topics

Resources

License

Contributing

Uh oh!

Stars

Watchers

Forks

Releases 1

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

2.3 Create `backend/.env.shared`

2.4 Create `backend/gateway/.env`

3.1 Create `frontend/.env.local`

`backend/.env.shared` (all workers)

`backend/gateway/.env` (gateway only)

`frontend/.env.local`

Packages