When should an agent use web search?

Use web search for unknown sources, one-off discovery, local long-tail facts, or coverage outside a monitored source set.

When should an agent use Synorb?

Use Synorb when the agent needs recurring source coverage, stable IDs, citations, provenance, Stream routing, and Manifests that can be stored or rendered.

Developer intent

Agent web search API, without making search do every job.

Q: Is Synorb an agent web search API?

Synorb is not a scrape-any-page search API. It is a source-grounded content feed and context graph for agents that need repeatable coverage, citations, and structured Manifests.

Agents need web search when they are discovering unknown pages. They need something different when the workflow depends on repeatable coverage: structured content objects, stable IDs, citations, source metadata, and a way to listen for new events.

Synorb fits that second path. It gives agents source-grounded Manifests from watched Streams, so production apps can use web search for gaps instead of making every agent run a fresh crawl.

Get Keys Search alternative →

Decision

Search is discovery. Feeds are operating context.

Use web search APIs for

Unknown source discovery

Good fitlong-tail facts, new pages, open-ended browsing, source discovery

Agent burdenrank, fetch, clean, judge credibility, dedupe, and remember staleness

Use Synorb for

Source-grounded feeds

Good fitmonitored topics, company feeds, research dashboards, cited app content

Agent burdenchoose Stream scope, pull Manifests, cite source URLs, store stable IDs

Response shape

Agents need objects they can trust and route.

An agent-ready search or feed result should not be just a snippet. It should include source URL, source name, published date, stable ID, summary, significance, tags, and a delivery contract the application can use.

{ "id": "manifest_id", "headline": "Source-grounded update", "summary": "What happened and why it matters.", "source_url": "https://source.example/story", "published_date": "2026-06-17", "stream_names": ["AI Infrastructure"], "tags": ["company", "product", "topic"] }

Implementation

Use Synorb first when the source universe is known.

During development

Use Synorb MCP in Cursor, Lovable, Claude, Codex, Windsurf, or another MCP client to discover Streams and inspect sample Manifests.

Build guides →

In production

Call Synorb REST from server-side code, keep credentials out of the browser, and render normalized rows with citations.

API docs →

For gaps

Use web search after the feed when the request falls outside monitored sources or needs ad hoc discovery.

Cost-aware pattern →

FAQ

Short answers.

Is Synorb an agent web search API?

No. Synorb is a source-grounded feed and context graph. It can replace many recurring agent web-search calls, but it is not a general crawl-any-page search API.

What does an agent get from Synorb?

Manifests with source URLs, summaries, Signals, Briefs, stable IDs, source metadata, tags, usage, and pagination.

Which Synorb interface should I use?

Use MCP while building with an agent. Use REST, webhooks, or S3 from your backend in production.

Start

Give the agent a feed before it searches.

Get Keys Real-time web data → Research monitor guide →