RonanCodes/hn-scan

HN Scan

Pull threads or topic results from Hacker News and return the distilled findings to the caller. No vault, no file writes — just a summary the caller can use.

Usage

/ro:hn-scan "agent observability"              # search front page + recent
/ro:hn-scan "show hn AI agent" --since 30d     # search Show HN in last 30 days
/ro:hn-scan https://news.ycombinator.com/item?id=47799856   # fetch one thread
/ro:hn-scan frontpage --limit 30               # top N current front-page stories
/ro:hn-scan show --since 7d                    # Show HN in last week
/ro:hn-scan ask --since 7d                     # Ask HN in last week

Modes

1. Topic search (free-text query)

Algolia HN search — no auth, no rate limit in practice.

curl -s "https://hn.algolia.com/api/v1/search?query=<encoded-query>&tags=story&hitsPerPage=30"
# Add &numericFilters=created_at_i>$(date -v-7d +%s) for --since 7d

For each hit return:

• title, points, num_comments, author, url (external link), story_id, created_at
• HN link: https://news.ycombinator.com/item?id={story_id}

Rank by points × recency (points / (age_hours + 2)^1.8).

2. URL mode (single thread)

If the arg is a HN URL or bare item id, fetch the full tree:

curl -s "https://hn.algolia.com/api/v1/items/{id}"

Return:

• Post metadata (title, points, author, date, linked URL)
• Top 10 comments by score (flatten, strip HTML to plain text, truncate to ~400 chars each)
• Total comment count

3. Frontpage / Show / Ask modes

# frontpage
curl -s "https://hn.algolia.com/api/v1/search?tags=front_page&hitsPerPage=30"
# Show HN
curl -s "https://hn.algolia.com/api/v1/search_by_date?tags=show_hn&hitsPerPage=30"
# Ask HN
curl -s "https://hn.algolia.com/api/v1/search_by_date?tags=ask_hn&hitsPerPage=30"

Output format

Return a terse report the caller can reason over. No markdown files, no YAML frontmatter.

# HN scan: "<query>"  (<N> results, last <window>)

1. [<points>pts, <comments>c] <title>
   <external-url-if-any>
   https://news.ycombinator.com/item?id=<id>
   <1-line takeaway if comment count > 50 and the story is relevant>

2. ...

For single-thread mode include the top-comments summary under the header.

HTML cleanup

Algolia returns HTML-encoded text. Before returning, convert:

• <p> → double newline
• <a href="X">Y</a> → [Y](X)
• <code> / <pre> → backticks / fences
• Common entities (', &, >, <, ") → literal chars
• Strip any remaining tags

Dependencies

None — uses curl + the free Algolia HN API.

See also

• llm-wiki/.claude/skills/ingest-hackernews — the sibling skill that persists full threads into an Obsidian-style vault. Use that when you want to keep the content long-term; use /ro:hn-scan when you just want to see what's there.
• /ro:trend-scan — orchestrator that runs hn-scan + x-scan + linkedin-scan + reddit-scan in parallel for a topic.
• /ro:x-scan, /ro:linkedin-scan, /ro:reddit-scan — siblings for other sources.