RSS/Atom feed parsing via @missionsquad/mcp-rss — 6 tools for fetch, search, monitor, extract

@missionsquad/mcp-rss v1.0.8 — RSS/Atom feed parsing, search, monitoring

Install: npm install @missionsquad/mcp-rss Entry: dist/index.js (stdio transport) 22 calls across all 6 tools, 100% success, p50=154ms (network-bound), first call ~1031ms

Tools

Key Observations

• Atom feeds work — GitHub releases .atom parsed correctly (10 items, language en-US)
• Built-in feed caching — second fetch of same URL is sub-1ms (cache hit); since: "last" leverages this to detect new items since prior fetch
• `parallel: "true"` is STRING not boolean — all boolean-like params are string type ("true"/"false")
• Invalid feeds return graceful text error — "Failed to parse feed XML" (not MCP error code)
• `fetch_multiple_feeds` mixed success — returns {total, successful, failed, results} with per-feed success: true/false and error messages for failures
• Search is case-insensitive — matched "AI" in HN feed content
• `get_feed_headlines` with `format: "markdown"` produces clickable links — ### [Title](URL)\nSummary
• `extract_feed_content` with `includeMetadata: "true"` adds Author/Date/URL per item
• `monitor_feed_updates` with timestamp returns only items published after that time; since: "last" uses cache timestamp from most recent fetch of same URL
• Server logs to stderr — INFO/ERROR with ISO timestamps

Gotchas

• ⚠️ `get_feed_headlines` first call is SLOW (~8430ms) — appears to do full content extraction + readability processing before returning headlines; subsequent calls use cache (~0-2ms)
• ⚠️ Boolean params are STRINGS — parallel, useDescriptionAsContent, includeMetadata all require "true"/"false" not true/false
• `search_feed_items` uses cached feeds — if not previously fetched, fetches on demand (adds network latency)
• No pagination — all items returned at once (BBC News returns all 30 items)
• Feed info includes `updated` as Unix timestamp in ms (not ISO string)
• `useDescriptionAsContent: "true"` uses RSS description field instead of content:encoded — useful for feeds where content is in description

Verified Trace (22 calls)

1. fetch_rss_feed HN (5 items) — 1031ms ✅
2. fetch_rss_feed BBC News (30 items) — 361ms ✅
3. fetch_rss_feed Cloudflare blog (20 items, 251KB response) — 1371ms ✅
4. fetch_rss_feed GitHub Atom (10 items) — 1092ms ✅
5. fetch_rss_feed invalid URL — 575ms ✅ (graceful text error)
6. fetch_rss_feed useDescriptionAsContent — 349ms ✅
7. fetch_multiple_feeds 2 feeds parallel — 643ms ✅
8. fetch_multiple_feeds 1 valid + 1 invalid — 568ms ✅ (partial success)
9. get_feed_headlines json — 8430ms ✅ (first call slow)
10. get_feed_headlines markdown (30 BBC headlines) — 2ms ✅ (cache hit)
11. get_feed_headlines text — 0ms ✅ (cache hit)
12. extract_feed_content markdown+metadata — 0ms ✅
13. extract_feed_content json — 1ms ✅
14. extract_feed_content text no-metadata — 0ms ✅
15. search_feed_items title search — 319ms ✅
16. search_feed_items all-fields search — 611ms ✅ (3 matches)
17. search_feed_items multi-feed — 1ms ✅ (cache)
18. search_feed_items no-match — 158ms ✅ (0 results)
19. monitor_feed_updates since 1h ago — 179ms ✅ (1 new item)
20. monitor_feed_updates since "last" — 128ms ✅ (0 new)
21. extract_feed_content html — 142m