telegraph

name	telegraph
description	Publish articles to Telegra.ph (Telegraph) using the createPage API via curl. Use when: user asks to publish, post, or create a page on Telegraph/Telegra.ph, or when a cron/task needs to publish formatted content to a Telegraph page. Also triggers for article publishing, blog posting to Telegraph, or generating readable web pages from content.
trigger	auto
keywords	["telegraph","telegra.ph","publish","article"]
metadata	{"openclaw":{"emoji":"📝","requires":{"bins":["curl"]}}}

Telegraph Publishing Skill

Publish formatted articles to Telegra.ph using the API.

When to Use

✅ USE this skill when:

• Publishing content to Telegra.ph (news reviews, summaries, reports)
• Creating a readable web page from compiled content
• A cron job needs to publish and share a link
• "Publie ça sur Telegraph" / "Create a Telegraph page"

❌ DON'T use this skill when:

• Reading or scraping existing Telegraph pages (just use curl directly)
• Managing a Telegraph account (editing, revoking tokens)

Authentication

The Telegraph access token is a plain-text file. Read it with read_file before calling the API.

Common token locations (check agent workspace or secrets directory):

• secrets/telegraph.txt
• workspace/telegraph-token.txt

Publishing a Page

API Call

curl -s -X POST "https://api.telegra.ph/createPage" \
  -H "Content-Type: application/json" \
  -d '{
    "access_token": "TOKEN",
    "title": "Your Page Title",
    "author_name": "Sclaw",
    "content": CONTENT_JSON,
    "return_content": false
  }'

Response

{
  "ok": true,
  "result": {
    "path": "Your-Page-Title-03-09",
    "url": "https://telegra.ph/Your-Page-Title-03-09",
    "title": "Your Page Title",
    "author_name": "Sclaw",
    "views": 0
  }
}

The published URL is in result.url.

Content Format

The content field is a JSON array of node objects. Each node has a tag and children.

Available Tags

Tag	Usage
h3	Section heading
h4	Sub-heading
p	Paragraph
strong	Bold text
em	Italic text
a	Link (href attribute)
blockquote	Quote block
ul, li	Bullet list
br	Line break
figure, img	Image

Node Structure

{"tag": "p", "children": ["Plain text content"]}

Nested elements:

{"tag": "p", "children": [{"tag": "strong", "children": ["Bold"]}, " then normal"]}

Links:

{"tag": "p", "children": [{"tag": "a", "attrs": {"href": "https://example.com"}, "children": ["Click here"]}]}

Example: Article with Sections

[
  {"tag": "p", "children": ["Introduction paragraph."]},
  {"tag": "h3", "children": ["🌍 Section Title"]},
  {"tag": "p", "children": ["Section content with details."]},
  {"tag": "h3", "children": ["💻 Another Section"]},
  {"tag": "p", "children": ["More content here."]},
  {"tag": "p", "children": [{"tag": "em", "children": ["Footer — Source attribution — Date"]}]}
]

Critical Rules for JSON Content

Telegraph content is sent as JSON inside a JSON body. This creates escaping challenges:

1. No curly/typographic quotes (' ' " ") — use straight ASCII quotes only
2. No unescaped apostrophes in French text — replace l'article with l article or rephrase (la suite de l article → cet article)
3. Escape inner double quotes — use \" inside JSON strings
4. Write the JSON to a temp file if the content is long, then use curl -d @/tmp/telegraph.json to avoid shell escaping issues

Recommended Pattern for Long Content

Instead of fighting shell escaping, write the request body to a file:

# 1. Build the JSON body and write to file (via write_file tool)
# 2. POST from file
curl -s -X POST "https://api.telegra.ph/createPage" \
  -H "Content-Type: application/json" \
  -d @/tmp/telegraph-request.json

This avoids all shell escaping problems.

Complete Workflow Example

Here's the typical flow for publishing a compiled article:

1. Gather content — fetch sources, compile topics
2. Read token — read_file secrets/telegraph.txt
3. Build request body — construct the JSON with title, author, and content nodes
4. Write to temp file — write_file /tmp/telegraph-request.json with the full request body
5. Publish — exec curl -s -X POST "https://api.telegra.ph/createPage" -H "Content-Type: application/json" -d @/tmp/telegraph-request.json
6. Extract URL — parse result.url from the response
7. Share — include the URL in the final message or notification