Publish Content to Telegraph with Media

This skill publishes content to Telegraph, the lightweight publishing platform, via its API, supporting images, YouTube embeds, PlantUML/Mermaid diagrams, and automatic splitting of long articles into linked multi-page sets. It's best suited for articles, research reports, documentation, and illustrated content. Rather than passing raw Markdown, it converts HTML fragments into Telegraph's required Node JSON format automatically, using a whitelist of supported tags (a, b, blockquote, code, figure, h3/h4, img, lists, p, pre, iframe, video, and more) - unsupported tags are stripped while their children are kept, and HTML tables specifically get converted into a readable monospace pre block since Telegraph has no native table node, with explicit guidance on when a table is more appropriate than a diagram (exact values matter) versus when a diagram is better (trends, flow, cohorts) and a mobile-first rule against wide dense tables that wrap badly on phones.

Its account model is API-only with no password or email: create_account.sh generates an access token plus a one-time auth URL to bind the API account to a browser session, after which pages are editable both via the API and at telegra.ph directly, with --revoke available to rotate a compromised token. For permanent media, the skill explicitly prefers a GitHub-backed workflow over Telegraph's own unofficial, unstable upload endpoint: images and diagrams are uploaded to a dedicated public GitHub assets repo, scoped to a fine-grained PAT with only Contents: Read and write on that one repo, and served via jsDelivr's CDN for stable permanent URLs, with a manifest tracking each page's uploaded assets by path and Git SHA so they can later be cleanly deleted via github_delete_page_assets.sh, keyed by Telegraph's stable path identifier rather than the page title, which can change. Telegraph's own upload.sh endpoint is explicitly demoted to a legacy fallback only, described as glitchy and unreliable behind corporate proxies/VPNs. Diagrams render through public PlantUML/Mermaid servers, a privacy consideration for confidential content, and can similarly be piped into the GitHub-backed permanent hosting flow. YouTube URLs are auto-normalized from watch?v=/youtu.be formats into embeddable iframe format. An optional integration with a separate fal-ai-image skill can generate illustrations before publishing, with a mandatory budget confirmation step (starting at $0.15/image) before generating anything.

Use this skill when publishing an article, research report, or illustrated document to Telegraph, especially when the content includes images, diagrams, tables, or YouTube video embeds, or when a long article needs automatic splitting across multiple linked pages (content over roughly 60KB triggers auto-split with an index page).

It requires a TELEGRAPH_ACCESS_TOKEN, created via create_account.sh if none exists, and for stable permanent media, a separate GitHub repo and scoped PAT - without that GitHub setup, media falls back to Telegraph's unreliable unofficial upload endpoint. It only accepts HTML fragments as input in this version, not raw Markdown, and diagram rendering sends source content to public third-party servers, so it should not be used for confidential diagrams.

Input is a page title plus content as an HTML string, an HTML file, or pre-built Node JSON, along with any local images/diagrams needing permanent hosting. Output is a published or edited Telegraph page, with media served via stable jsDelivr URLs when GitHub-backed hosting is used, long content automatically split into an indexed multi-page set, and a per-page manifest tracking uploaded assets for later cleanup.

sh scripts/create_page.sh --title "Article" --html-file article.html

Built on the Telegraph API, a GitHub assets repo plus jsDelivr CDN for permanent media hosting, public PlantUML/Mermaid rendering servers for diagrams, and an optional fal-ai-image skill integration for generating illustrations before publishing.

Writers and researchers publishing illustrated articles, reports, or documentation to Telegraph who want reliable permanent media hosting, automatic long-article splitting, and clean diagram/table/video embedding rather than manually wrangling Telegraph's raw API and unstable upload endpoint.