community-base

Shared Django apps intended for two DataTalks.Club community sites:

• DataTalks.Club website (DataTalksClub/website)
• AI Shipping Labs website (AI-Shipping-Labs/website)

Both sites run independently, with their own database, deployment, user model, public design and site-specific apps. They install this package and mount the apps they need. Email, contacts, campaigns, background jobs and schedules are delegated to Relay (DataTalksClub/relay), a separate service that each site talks to over HTTPS.

Status: Phase 0 in progress. The package skeleton and first shared primitives are being built; the plan in docs/plan/ moves code out of the two sites phase by phase.

Start here

• Agents: read AGENTS.md (entry point), then follow docs/PROCESS.md for each issue.
• Progress across all four repositories: docs/plan/STATUS.md. python scripts/plan.py next prints what can start now; python scripts/plan.py summary prints totals.
• Work in the site repositories follows their own process (AGENTS.md and _docs/PROCESS.md in AI-Shipping-Labs/website and DataTalksClub/website, docs/PROCESS.md in DataTalksClub/relay). This plan says what to build and how to verify it; the site process says how work is done there.

Documents

Document	What it is for
AGENTS.md	Entry point for agents: reading order, how to pick and do an issue, tracking rules, site-process rule.
docs/PROCESS.md	The per-issue workflow with a check at every step.
docs/plan/STATUS.md	Progress tracker for every issue in every repository. Generated by scripts/plan.py.
docs/00-analysis.md	Inventory of the two sites and Relay, overlap map, coupling audit. Read once.
docs/01-decisions.md	Product and architecture decisions already taken by the owner. Do not re-open them.
docs/02-architecture.md	Target architecture, package layout, app labels, extension seams, template contract.
docs/03-playbooks.md	Step-by-step procedures reused by many issues: lifting an app, squashing migrations, cutting a seam, releasing, freeze weekend.
docs/04-quality-gates.md	The checks every issue must pass, the issue template, and when to stop and ask a human.
docs/plan/README.md	Phase index and dependency order.
docs/plan/phase-*.md	One file per phase with every issue spelled out: goal, what to read, steps, verification, done criteria.

How to work on this

1. Read docs/04-quality-gates.md first. Every issue in docs/plan/ assumes those gates.
2. Pick the lowest-numbered open issue in the current phase whose dependencies are done.
3. Follow the issue steps literally. Run every verification command and compare with the expected result. If a check fails, fix it before moving on. Never skip a check.
4. Open one pull request per issue, in the repository the issue names (community-base, DataTalksClub/website, AI-Shipping-Labs/website, or DataTalksClub/relay), following that repository's process.
5. Update docs/plan/STATUS.md when you start and when you finish.

Conventions

• Python is managed with uv. Never call pip directly.
• Documents use plain text, headings, tables and backticks. No bold formatting.
• File paths, commands, setting keys and model names are written in backticks.
• Site repositories keep their own process documents; this repository's per-issue process is docs/PROCESS.md and its gates are docs/04-quality-gates.md.