rails-agents CLI

sync

Push every file under app/agents/<name>/ (plus resolved Library imports from imports.yml) to Rails Agent Cloud. Prefer sync over any deprecated cloud-pull authoring flow.

$ bundle exec rails-agents sync store_assistant
# Synced N files for store_assistant
# (includes app/agents_library attachments when imports.yml is present)

Sync fails fast on Library path traversal, missing sources, or conflicts with agent-local files — so the Library stays canonical without duplicating shared source in Git.

run

Execute the agent against the cloud runtime and print output_text. Pass --session for conversation memory continuity.

$ bundle exec rails-agents run store_assistant \
    "Where is order ORD-DEMO-1001?"

$ bundle exec rails-agents run store_assistant \
    "Remind me what we discussed" --session user-42

run is mapped to the underlying execute Thor task. Integrate the same result shape in product code via run response schema.

deploy

Deploy the agent bundle for production (or named) traffic. Omit the name to deploy every discovered agent under app/agents/.

$ bundle exec rails-agents deploy store_assistant
$ bundle exec rails-agents deploy   # all agents

Typical loop: edit → sync → run (or dashboard Test) → evals → deploy.

logs

Fetch recent run logs from the cloud as JSON. Optional agent filter and limit.

$ bundle exec rails-agents logs store_assistant
$ bundle exec rails-agents logs store_assistant --limit 20
$ bundle exec rails-agents logs   # workspace-wide

Related: bundle exec rails-agents traces [NAME] for tool-level traces. The Monitor UI in /agents remains the primary day-to-day surface.

evals

Fetch eval results recorded for an agent (or the workspace). Define cases locally under evals/*.yml, sync them, and run from deploy / dashboard when you want a regression gate.

$ bundle exec rails-agents evals store_assistant
$ bundle exec rails-agents evals

Also useful