Be the first to get access to new connectors. Get early access →
ConnectorBay Early access

Daybrief — connector documentation

  • MCP endpoint: https://daybrief.connectorbay.com/mcp
  • REST API: POST https://daybrief.connectorbay.com/v1/tools/{tool_name}, described by the OpenAPI 3.1 spec at https://daybrief.connectorbay.com/openapi.json
  • Transport: Streamable HTTP, stateless, MCP specification 2025-06-18 or later
  • Authentication: none required. An optional Authorization: Bearer <key> raises rate limits.
  • Health: https://daybrief.connectorbay.com/health · Tool schemas: https://daybrief.connectorbay.com/tools.json
  • Publisher: Unify Dynamics LLC · connectorbay.com/daybrief · agent@connectorbay.com

What Daybrief is for

A news app shows one outlet's feed. Daybrief shows the story: the same event as seen by NPR, the BBC, the New York Times, the Guardian, CBS, NBC, PBS, Al Jazeera, CNN, the Washington Post, Politico, MarketWatch, Ars Technica, The Verge, TechCrunch, WIRED, MIT Technology Review, STAT, ScienceDaily, Nature, ESPN, CBS Sports and Variety — clustered, ranked by how many outlets carry it, with a one-line summary and a link to each outlet's version. On top of that: a personal briefing with interests and muted topics, "what did I miss" sized to a few minutes, worldwide search and day-by-day coverage timelines (GDELT), Wikipedia background and the editor-curated current-events digest, the Hacker News front page, and context for any link a user pastes.

What it never does

Daybrief reads the RSS/Atom feeds outlets publish for exactly this purpose — headlines, short summaries (capped at 280 characters) and links — and never fetches, stores or returns article bodies. Every story carries the outlet name and a link to read there. For a pasted link, it reads only the page's own meta tags, as any link preview does. Nothing about the user is stored.

Tools

Digest

top_stories — The biggest stories in a section right now, clustered across outlets: one headline per story, how many outlets are covering it, a one-line summary and links to each outlet's version. Use for "what's the news", "what's happening in tech/business/science", and as the first call for any digest. Args: section, hours, limit, mute.

briefing — A personal news briefing in one call: the top clustered stories for each interest (sections or free-text topics), with muted topics removed, deduplicated across interests, and a reading-time estimate. Use for "catch me up", "morning briefing", "what did I miss", "news but nothing about X". Args: interests, mute, hours, per_interest, include_tech_pulse.

most_covered — The stories the most outlets are covering right now, across every section: the cross-outlet consensus of what matters today. Use for "what's the biggest story", "what's everyone talking about". Args: hours, limit, mute.

catch_up — "What did I miss?" — the most-covered stories published since the user was last looking, optionally focused on their interests and with muted topics removed, sized to read in a couple of minutes. Args: since_hours, interests, mute, limit.

Find & trace

search_news — Find coverage of a topic: first across the curated publisher feeds (clustered), then, if requested, GDELT's worldwide index for outlets we don't carry. Use for "any news about X", "what's the latest on Y". Args: query, hours, limit, worldwide.

story_timeline — How coverage of a topic rose and fell day by day (share of all monitored global news, from GDELT), plus the peak day. Use for "is this story growing or fading", "when did this blow up". Args: query, days.

coverage — Who is covering a story: which of the curated outlets have it (with their headlines), how many, and, when GDELT is reachable, which countries' press are covering it. Use for "is this being widely reported", "how is X being covered abroad". Args: query, hours.

local_news — Recent coverage mentioning a place, from the curated feeds and GDELT's index of local outlets. Use for "what's happening in ". Note: national feeds rarely cover small towns; GDELT does. Args: location, hours, limit.

outlet_feed — The latest headlines from one outlet across the sections we carry for it. Use when the user names a source. Args: outlet, limit, mute.

Context

whats_happening — Wikipedia's editor-curated current-events digest for a day: armed conflicts, disasters, politics, science, sports and more, each item with source links and the Wikipedia topics involved. A calm, sourced overview; use alongside top_stories for context and for "what happened on ". Args: day.

background — Wikipedia's summary of a topic plus, optionally, the current stories mentioning it: the "who/what is this" behind a headline. Use for "explain the background", "who is X". Args: topic, with_news.

story_context — Context for a link someone shared: the page's own title, summary, outlet, author and publish time (from its meta tags, like any link preview), plus other outlets' coverage of the same story from the curated feeds and Wikipedia background on the main topic. Use when a user pastes a news link and asks "is this true / what's this about / who else is reporting it". Args: url.

tech_pulse — What builders are reading: the Hacker News front page with points, comment counts, source domains and discussion links. Use for "what's on HN", "what's the tech community talking about", startup and developer news. Args: limit, ranking.

outlets — The outlets and sections Daybrief carries, with feed ids, for outlet_feed and to explain coverage. Args: .

Sections

top, world, us, politics, business, tech, science, health, sports, entertainment. Free-text interests in briefing are matched against every feed's headlines and summaries.

Response conventions

  • ok: true with the payload, or ok: false with error.code in bad_input, not_found, no_data, upstream_error.
  • Every story cluster: headline, snippet, lead_outlet, lead_link, published (UTC), outlet_count, outlets, coverage (each outlet's headline and link). outlet_count is the best signal of importance.
  • degraded: [...] names optional sources that were throttled (gdelt, wikipedia, hackernews) or feeds that timed out (feed:<id>). Clustering from the remaining feeds still works. GDELT allows one query every five seconds, so worldwide search and timelines can be unavailable under load; the curated results are unaffected.
  • reading_minutes estimates time to read the returned headlines and summaries.

Rate limits

Free tier: 60 tool calls per minute and 300 per day per client (agent user id, client id header, or IP). A Pro key raises this to 300/min and 5,000/day. Over the limit returns HTTP 429 with Retry-After. initialize and tools/list are never counted.

Sources and licences

Publisher RSS/Atom feeds (each outlet's own terms; used for syndication of headlines and links) · GDELT Project (open) · Wikipedia (CC BY-SA 4.0) · Hacker News API (official, MIT-licensed client terms).