Getting Started
Onboarding Devices
First, prepare the full Runtime distribution environment for the matching platform, including easynet, daemon, signing helper, and native bridge; when interactive terminals are needed, session supervisor is also required. The following assumes you already have usable accounts, Realm resolvable, Hub authentication/pairing API accessible, and daemon TLS endpoint reachable; private deployments require corresponding trust materials. Desktop collection also requires native media support, real devices, and system permissions.
Replace the account and Realm with your own values. To clearly see each stage, explicitly disable auto-join first:
easynet login --user YOUR_LOGIN --realm YOUR_REALM --no-join
easynet join --boot no
easynet runtime start
easynet runtime status
easynet device list --state all --format json
easynet ability listIt is expected to first save the account Profile, then establish device-member relationships, and finally start the Runtime. runtime status is used to check persistent snapshots and local lifecycles; device list proactively requests the live directory, and the nodes in the JSON are used to verify the current directory view. Currently login will by default continue to join and start; if joining fails you can recover with easynet join --profile YOUR_PROFILE. The ability list shows the projects actually connected in your environment; your own business functions need to start and publish a Provider first.
Map materials and connect to MCP
Context can be experienced even earlier: the local add/list does not require an account or daemon; in a project directory prepared for public browsing, you can run the first two commands. Agent access to these materials requires pairing. The MCP checks and configuration below reuse the prepared Runtime:
easynet context add . --name research
easynet context list
easynet context clipboard status
easynet mcp status
easynet mcp install codex --name easynet --dry-runDirectory mappings should show normalized paths; MCP status should report that the local daemon is reachable. After previewing the configuration, run the commands without --dry-run and restart Codex inspection tools. The installer obtains the tenant from the Runtime status; if the GUI environment cannot find the bridge, you can specify the library path with --bridge-lib to an absolute path.
Understanding task dependencies with a complete program
EAL can be experienced offline first: all you need is a runnable easynet CLI, no account, pairing, daemon, Hub, or real data services. Save the complete content below as pipeline.eal; it is identical in program body to the one in the repository at examples/pipeline.eal:
mission "etl-pipeline" {
let raw = call "data.extract" on "source-db" with {
query = "SELECT * FROM events WHERE ts > now() - interval '1 hour'"
} timeout 60
let cleaned = call "data.transform" on "compute-node" with {
input = raw.output,
rules = "deduplicate,normalize"
} timeout 120
let report = call "data.load" on "warehouse" with {
input = cleaned.output,
table = "events_cleaned"
} timeout 60
}In the directory where you saved the file, execute:
easynet mission compile pipeline.eal --emit-irExpected Mission IR JSON output: the mission name is etl-pipeline, containing three steps and three stages. Stage 0 extracts data and binds to raw; Stage 1's input references raw.output, and after completion binds cleaned; Stage 2 references cleaned.output. The timeouts are 60, 120, and 60 seconds respectively. The dependencies among inputs generate the execution order, which is exactly the orchestration capability this example aims to demonstrate.
These capability names and nodes are teaching placeholders. The compile step only parses and plans; it does not parse online targets, does not query databases, and will not produce an execution trace or run_id. Therefore you can verify data flow even without network access; a successful compile means the program structure is valid, not that all three targets exist.
To verify actual invocation, proceed to run the EasyRemote quotation case. After that, read Python Pipeline construction and Runtime execution distribution to organize already-registered capabilities into multi-step tasks. "}
No comments yet

