Freshrss category
Summarize recent posts from a named FreshRSS category. For the "People" category, writes a person-centric narrative (one paragraph per person); for other categories, writes one paragraph per active feed. Use when the user asks "what's new in my <category> feeds?", "catch me up on <category>", or invokes `/freshrss-category <category>`. Requires the freshrss MCP server.From its SKILL.md
npx -y skills add edsu/freshrss-claude --skill freshrss-categoryAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
4.6 KB, as published. Nobody here has run it
Category digest
Summarize recent unread posts from a specific FreshRSS category. For the People category, the output is person-centric: one paragraph per active person, written like a catch-up with someone you follow. For any other category, write one paragraph per active feed — what's the thread or mood of that feed this week?
When to use
- "What's new in my People feeds?"
- "Catch me up on Tech"
- Invoked as
/freshrss-category <category>(with optional timeframe)
Arguments
<category> [timeframe]
category— the FreshRSS category name (case-sensitive, e.g.People,Tech,News)timeframe— optional, defaults to 7 days (e.g.1d,3d,2w,last month)
Instructions
1. Parse arguments and compute cutoff timestamp
Extract category (first token) and timeframe (remainder, default 7d) from
args.
# macOS — substitute the parsed timeframe, e.g. for 7d:
date -v-7d +%s
# Linux
date -d '7 days ago' +%s
2. Fetch the feeds list
If mcp__freshrss__list_feeds schema isn't loaded:
ToolSearch with select:mcp__freshrss__list_feeds.
Call it (no arguments). The result is usually large and will be saved to a file — note the path. This gives you feed names with their FreshRSS categories.
3. Fetch unread articles
If mcp__freshrss__get_unread_articles schema isn't loaded:
ToolSearch with select:mcp__freshrss__get_unread_articles.
Call with:
since_timestamp: cutoff from step 1limit: 2000max_summary_length: 300
Note the saved file path.
4. Filter to the target category
python skills/freshrss-digest/scripts/process_articles.py \
"$articles_path" "$cutoff" \
--category "$category" \
--feeds-json "$feeds_path" \
> /tmp/freshrss_category.json
The script prints a per-feed post count survey to stderr — check it to see who's been active before writing.
4b. Print the stats header
python skills/freshrss-digest/scripts/stats_header.py /tmp/freshrss_category.json "category · {category} · {timeframe}"
Include this output verbatim at the top of your response.
5. Read the articles
python skills/freshrss-category/scripts/show_feeds.py /tmp/freshrss_category.json
This groups articles by feed with cleaned-up display names and full summaries. Read through all of it before writing.
6. Write the summary
If category is People: one paragraph per active person, ordered by post
count (most active first).
- Lead with the person's name (use the clean display name from the script).
- Synthesize across their posts — what's the thread or mood?
- Inline-link to specific posts naturally within the prose.
- Tone: warm and conversational — this is a catch-up, not a news brief.
- If someone only posted a link or a short reaction, say what it was about and why it seemed to matter to them.
For any other category: one paragraph per active feed, ordered by post count.
- Lead with the feed name.
- Summarize the week's posts for that feed — what topics, what angle?
- Inline-link to notable posts.
- Tone can be more neutral/informational than for People.
In both cases:
- Skip feeds/people with no posts in the window without mentioning them.
- Do not group by topic across feeds — keep each feed/person together.
7. Offer mark-as-read
After the summary, ask the user if they'd like to mark all articles in the
category as read. Load mcp__freshrss__mark_as_read via ToolSearch if needed,
then call it with the article IDs from the filtered JSON:
import json
articles = json.load(open("/tmp/freshrss_category.json"))
article_ids = [a["id"] for a in articles]
The tool returns {"ok": true} on success; confirm to the user.
Notes
- Category names are case-sensitive — use the exact name as it appears in
FreshRSS (e.g.
People, notpeople). process_articles.py --categoryrequires--feeds-jsonpointing at the savedlist_feedsoutput.- If the feeds list came back inline (rare, small subscriber count), save it to a temp file and pass that path.
- For a broader digest across all feeds, use
/freshrss-digest. - For searching within a category, use
/freshrss-search.