Miniflux News

Use the bundled script to fetch entries, then format a clean list and optionally write summaries.

Setup (credentials)

This skill reads Miniflux credentials from a local config file by default.

Config file (recommended)

Path:

• - INLINECODE0

Format:
CODEBLOCK0

Create/update it using the script:

CODEBLOCK1

Environment variables (override)

You can override the config file (useful for CI):

CODEBLOCK2

Token scope: Miniflux API token with read access.

Fetch latest entries

List latest unread items (default):

CODEBLOCK3

Filter by category (by name):

CODEBLOCK4

If you need machine-readable output:

CODEBLOCK5

Response formatting

• - Return a tight bullet list: [id] title — feed + link.
• Ask how many the user wants summarized (e.g., “summarize 3” or “summarize ids 123,124”).

View full content

Show the full article content stored in Miniflux (useful for reading or for better summaries):

CODEBLOCK6

If you want the raw HTML as stored by Miniflux:

CODEBLOCK7

Categories

List categories:

CODEBLOCK8

Mark entries as read (explicit only)

This skill must never mark anything as read implicitly. Only do it when the user explicitly asks to mark specific ids as read.

Mark specific ids as read:

CODEBLOCK9

Mark all unread entries in a category as read (still explicit, requires --confirm; includes a safety --limit):

CODEBLOCK10

Summarize entries

Fetch full content for a specific entry id (machine-readable):

CODEBLOCK11

Summarization rules:

• - Prefer 3–6 bullets max.
• Lead with the “so what” in 1 sentence.
• If content is empty or truncated, say so and summarize from title + available snippet.
• Don’t invent facts; quote key numbers/names if present.

Troubleshooting

• - If the script says missing credentials: set MINIFLUX_URL/MINIFLUX_TOKEN or create ~/.config/clawdbot/miniflux-news.json.
• If you get HTTP 401: token is wrong/expired.
• If you get HTTP 404: base URL is wrong (should be the Miniflux web root).