MCP Resources

the litmus mcp server exposes a set of litmus\ //docs/\<section> uris as mcp resources mcp aware clients can read these uris on demand, fetching live content directly from docs litmus io https //docs litmus io this means an llm client can pull current litmus reference material into its context without you copying and pasting documentation why resources mcp tools are for actions the model takes mcp resources are for static (or live fetched) reference material the model reads by exposing litmus docs as resources, an mcp client can pull driver specific documentation when configuring a new device read the devicehub or digital twins user guide when the model is unsure how a feature works confirm the current behavior of a litmus edge component before suggesting changes available resources uri source page litmus\ //docs/overview docs litmus io https //docs litmus io litmus\ //docs/edge /litmusedge litmus\ //docs/edge/devicehub /litmusedge/product features/devicehub litmus\ //docs/edge/digitaltwins /litmusedge/product features/digital twins litmus\ //docs/edge/datahub /litmusedge/product features/datahub litmus\ //docs/edge/marketplace /litmusedge/product features/applications litmus\ //docs/edgemanager /edgemanager litmus\ //docs/edgemanager/marketplace /edgemanager/lem user ui/product features/marketplace catalogs and applications litmus\ //docs/edgemanager/grafana /edgemanager/lem user ui/product features/grafana dashboards litmus\ //docs/solutions /solutions the full list is defined in src/tools/resource tools py how a client uses them standard mcp protocol client calls list resources > server returns the uris above client calls read resource(uri) with one of them > server fetches the page via httpx , returns the rendered text different mcp clients surface resources differently some expose them via a context menu ("attach resource"), some let the model autonomously call read resource mid conversation, some require manual selection live vs cached each read resource call performs a fresh httpx get against docs litmus io no caching, no bundling if the docs are updated, the next read picks up the changes see also tools reference # for the action oriented half of the mcp surface