It's Already Written.

The rules are already written.

A fully automated, self-publishing TTRPG publication. It scrapes, writes, publishes, and posts to X — no manual steps required.

Live site: https://bizzal70.github.io/itsalreadywritten
X: @ItsAlrdyWritten

Sister site to It's Already When. (cyber) and It's Already Priced. (crypto).

---

Three sections, three cadences

Section	Cadence	What it is
Issues	Weekly, Friday 10am MT	Long-form digest: releases, rules discourse, industry news, the week's verdict
Field Notes	Daily, 7am MT	Short tactical entries on the most relevant TTRPG item from the last 24 hours
RTFM	Weekly, Wednesday 9am MT	Evergreen rules deep dives grounded in official system documentation — not news-driven

Each section has its own index page and RSS feed.

---

How it works

7:00 AM MT daily          7:00 AM MT daily         Friday 10 AM MT          Wednesday 9 AM MT
─────────────────         ─────────────────         ───────────────          ─────────────────
RSS feeds (TTRPG)         Last 24h articles          Week's articles          Next unused topic
      │                         │                         │                         │
      ▼                         ▼                         ▼                         ▼
  scraper.py              field_note.py              issues.py                  rtfm.py
      │                         │                         │                         │
      ▼                         ▼                         ▼                         ▼
 articles.db              _posts/ field note          _posts/ issue            _posts/ rtfm
                                │                         │                         │
                                └─────────────────────────┴─────────────────────────┘
                                                          │
                                                          ▼
                                              deploy.yml → GitHub Pages
                                                          │
                                                          ▼
                                          X card thumbnail generated (Pillow)
                                                          │
                                                          ▼
                                                  Tweet posted to X

Everything runs on GitHub Actions. No local machine, server, or cron required.

---

Workflows

Workflow	Schedule	What it does
daily-field-note.yml	7am MT daily	Scrapes feeds; generates Field Note if anything is high-signal; pushes; tweets with card
weekly-issue.yml	Friday 10am MT	Catches up missed articles; generates Issue with Claude; pushes; tweets with card
weekly-rtfm.yml	Wednesday 9am MT	Picks next unused RTFM topic; generates article; pushes; tweets
tweet-on-publish.yml	On push to _posts/	Waits for page to go live (200 OK), then posts tweet with generated thumbnail card
deploy.yml	On every push to main	Builds + deploys Jekyll site to GitHub Pages

All scheduled workflows support workflow_dispatch for manual runs.

---

Scraper components

File	Purpose
scraper/feeds.py	TTRPG sources: 4 RSS feeds (EN World, Sly Flourish, Gnome Stew, Tribality) plus dice.camp Mastodon tag feeds (the TTRPG Mastodon)
scraper/scraper.py	Pulls feeds, deduplicates by URL hash, caches to articles.db
scraper/issues.py	Generates weekly Issue from this week's unused articles via Claude API
scraper/field_note.py	Generates daily Field Note from last 24h articles; skips if nothing high-signal
scraper/note_quality.py	Deterministic editorial substance floor for Field Notes: asserts word count / structure / concrete specifics. A thin note is regenerated once (stricter), then skipped
scraper/rtfm.py	Picks next unused topic from topic backlog; generates evergreen RTFM article
scraper/sources.py	Curated, stable per-system source URLs (D&D 5e SRD, Shadowdark, DCC) — never deep-links or LLM-generated URLs
scraper/resources.py	Builds deterministic source citations for rules references
scraper/registries.py	Tracks published issues and field notes to prevent duplicates
scraper/style.py	Per-system tone and voice rules for Claude prompts
scraper/x_thumbnail.py	Generates 1200×675 X card image (Pillow): warm dark background, amber accent, blog name + issue number + summary
scraper/tweet_on_publish.py	Posts tweet with thumbnail when a new post is detected; waits for page to go live first

---

Content rules

• Headline-first titles — the generator asks the model for a HEADLINE: line and promotes it to the post title; the Issue #NNN · Week of … label drops to a kicker line above the body. The title slot does click-work instead of filing the post. Falls back to the summary, then the old label, if no headline comes back
• Every post closes with a follow CTA — weekly Issues and daily Field Notes both end with a follow/subscribe line (@ItsAlrdyWritten or RSS), so a post is never a dead end
• No em dashes — a deliberate choice to avoid an obvious AI-writing tell
• No AI disclosure in post copy
• RTFM is not news-driven — grounded only in official system documentation, not the scraped article DB
• Source links are deterministic — sources.py uses curated, stable landing pages (SRD pages, publisher sites); never hallucinated deep links
• Per-system tone — style.py controls voice per RPG system (D&D 5e, Shadowdark, DCC)
• Duplicate prevention — registries.py tracks published items so re-runs never double-post

---

Required secrets

Settings → Secrets and variables → Actions:

Secret	Purpose
GH_PAT	Personal access token with repo write access (used for committing posts)
ANTHROPIC_API_KEY	Claude API for content generation
X_API_KEY / X_API_SECRET	X app consumer keys (OAuth 1.0a)
X_ACCESS_TOKEN / X_ACCESS_TOKEN_SECRET	X user access tokens

---

Site structure

_posts/           Issues + Field Notes + RTFM articles (all in one collection)
_layouts/         Jekyll templates (post, default)
assets/           CSS, fonts, images
scraper/          All automation scripts
  state/          Published item registries (committed, version-controlled)
.github/workflows/ GitHub Actions workflows

Issue numbers tracked in scraper/issue_number.txt, incremented on each Issue run.

License

• Code (generators, workflows, templates, config): MIT
• Written content (posts, articles, field notes): CC BY-NC 4.0 — attribution required, non-commercial reuse only.