Research and Server
ResearchConfig::new(topic) supports depth, maximum sources, verification, and engine settings.
Workflow and current contract
ResearchConfig::new(topic) supports depth, maximum sources, verification, and engine settings. research runs the configured pipeline; research_with_engine allows an explicit research engine. Inspect returned sources and failures when using a report in an agent workflow.
The HTTP router exposes POST /fetch, /observe, /search, /perceive, and /research, plus GET /health. WEAVE_API_KEY enables request authentication. The server applies a 1 MiB request-body limit and a 120-second request timeout; serve only warns about a non-loopback bind when the key is empty and still starts the listener. The keyless authentication branch allows requests, so callers must configure the key or explicitly bind to loopback.
Keep the server future alive by awaiting serve(addr). Spawning it immediately before returning from main does not keep a Tokio process running.
Source reference
researchdeclarations —libs/weave-browser-engine/src/research.rs.servedeclarations —libs/weave-browser-engine/src/serve.rs.
