Ruby SDK for the Agent Runtime Control Protocol (ARCP) — submit, observe, and control long-running agent jobs from Ruby.
Specification · Concepts · Install · Quick start · Guides · API reference
arcp is the Ruby reference implementation of ARCP, the Agent Runtime Control Protocol. It covers both sides of the wire — Arcp::Client for submitting and observing jobs, Arcp::Runtime::Runtime for hosting agents — so either side can talk to any conformant peer in any language without hand-rolling the envelope, sequencing, or lease enforcement.
ARCP itself is a transport-agnostic wire protocol for long-running AI agent jobs. It owns the parts of agent infrastructure that don't change between products — sessions, durable event streams, capability leases, budgets, resume — and stays out of the parts that do. ARCP wraps the agent function; it does not define how agents are built, how tools are exposed (that's MCP), or how telemetry is exported (that's OpenTelemetry).
Requires Ruby 3.3 or later. The gem runs on the socketry/async reactor and pulls in async-websocket for the default networked transport and sqlite3 for the resume log; no separate extras are needed. Add it to a Gemfile:
gem 'arcp', '~> 1.0'bundle installConnect to a runtime, submit a job, stream its events to completion:
require 'async'
require 'arcp'
ECHO = lambda do |ctx|
ctx.log(level: 'info', message: "echoing #{ctx.input.inspect}")
ctx.progress(current: 1, total: 1, units: 'message')
ctx.finish(result: { 'echoed' => ctx.input })
end
Sync do
runtime = Arcp::Runtime::Runtime.new(
auth_verifier: Arcp::Auth::Bearer.from_token('demo', principal_id: 'alice'),
heartbeat_interval_sec: nil
)
runtime.register_agent(name: 'echo', versions: ['1.0.0'], default: '1.0.0', handler: ECHO)
server_t, client_t = Arcp::Transport::MemoryTransport.pair
server = Async { runtime.accept(server_t) }
client = Arcp::Client.open(
transport: client_t,
auth: { 'scheme' => 'bearer', 'token' => 'demo' },
client_name: 'quickstart'
)
handle = client.submit_job(agent: 'echo', input: { 'msg' => 'hi' })
handle.subscribe(client: client).each { |event| puts "#{event.kind}: #{event.body.to_h}" }
result = handle.get_result(client: client)
puts "final: #{result.final_status} #{result.result.inspect}"
client.close
server.stop
endThis is the whole shape of the SDK: open a session, submit work, consume an ordered event stream, get a terminal result or error. Everything below is detail on those four moves.
ARCP organizes everything around four concerns — identity, durability, authority, and observability — expressed through five core objects:
- Session — a connection between a client and a runtime. A session carries identity (a bearer token), negotiates a feature set in a
hello/welcomehandshake, and is resumable: if the transport drops, you reconnect with a resume token and the runtime replays buffered events. Jobs outlive the session that started them. See §6. - Job — one unit of agent work submitted into a session. A job has an identity, an optional idempotency key, a resolved agent version, and a lifecycle that ends in exactly one terminal state:
success,error,cancelled, ortimed_out. See §7. - Event — the ordered, session-scoped stream a job emits: logs, thoughts, tool calls and results, status, metrics, artifact references, progress, and streamed result chunks. Events carry strictly monotonic sequence numbers so the stream survives reconnects gap-free. See §8.
- Lease — the authority a job runs under, expressed as capability grants (
fs.read,fs.write,net.fetch,tool.call,agent.delegate,cost.budget,model.use). The runtime enforces the lease at every operation boundary; a job can never act outside it. Leases may carry a budget and an expiry, and may be subset and handed to sub-agents via delegation. See §9. - Subscription — read-only attachment to a job started elsewhere (e.g. a dashboard watching a job a CLI submitted). A subscriber observes the live event stream but cannot cancel or mutate the job. Distinct from resume, which continues the original session and carries cancel authority. See §7.6.
The SDK models each of these as first-class objects; the rest of this README shows how.
Open a session, negotiate features, and reconnect transparently after a transport drop using the resume token — jobs keep running server-side while you're gone.
require 'async'
require 'arcp'
Sync do
client = Arcp::Client.open(
transport: transport,
auth: { 'scheme' => 'bearer', 'token' => ENV.fetch('ARCP_TOKEN') },
client_name: 'resumable'
)
session_id = client.session.id
resume_token = client.session.resume_token
last_seq = Hash.new(0)
handle = client.submit_job(agent: 'long-runner')
handle.subscribe(client: client).each do |event|
last_seq[handle.job_id] = event.body.respond_to?(:seq) ? event.body.seq : last_seq[handle.job_id]
end
# ... transport drops ...
resumed = Arcp::Client.open(
transport: new_transport,
auth: { 'scheme' => 'bearer', 'token' => ENV.fetch('ARCP_TOKEN') },
resume: { 'token' => resume_token, 'last_event_seq' => last_seq }
)
# The runtime replays every event with event_seq > last_seq, then resumes live streaming.
endSubmit a job with an agent (optionally version-pinned as name@version), an input, and an optional lease request, idempotency key, and runtime limit.
handle = client.submit_job(
agent: 'weekly-report@2.1.0',
input: { 'week' => '2026-W19' },
lease_request: Arcp::Lease::LeaseRequest.new(
capabilities: ['net.fetch'],
expires_at: (Time.now.utc + 60).iso8601
),
lease_constraints: Arcp::Lease::LeaseConstraints.new(
expires_at: (Time.now.utc + 300).iso8601,
max_budget: nil
),
idempotency_key: 'weekly-report-2026-W19',
max_runtime_sec: 300
)
puts "job_id = #{handle.job_id}"
puts "resolved agent = #{handle.agent.inspect}"
puts "effective lease = #{handle.lease&.to_h.inspect}"Iterate the ordered event stream — log, thought, tool_call, tool_result, status, metric, artifact_ref, progress, result_chunk — and optionally acknowledge progress so the runtime can release buffered events early.
last_seq = 0
handle.subscribe(client: client).each do |event|
case event.kind
when Arcp::Job::EventKind::LOG
puts event.body.message
when Arcp::Job::EventKind::TOOL_CALL
puts "-> tool #{event.body.name}(#{event.body.arguments.inspect})"
when Arcp::Job::EventKind::METRIC
puts "metric #{event.body.name}=#{event.body.value}#{event.body.unit}"
when Arcp::Job::EventKind::PROGRESS
puts "progress #{event.body.current}/#{event.body.total} #{event.body.units}"
end
last_seq += 1
client.ack(last_seq) if (last_seq % 32).zero? # coalesced session.ack
endRequest capabilities, a budget, and an expiry; read budget-remaining metrics as they arrive; handle the runtime's enforcement decisions.
handle = client.submit_job(
agent: 'web-research',
input: { 'iterations' => 8 },
lease_request: Arcp::Lease::LeaseRequest.new(
capabilities: ['tool.call', 'cost.spend'],
budget: Arcp::Lease::CostBudget.parse(['USD:1.00']),
expires_at: (Time.now.utc + 600).iso8601
)
)
puts "initial budget = #{handle.lease&.budget&.to_a.inspect}"
handle.subscribe(client: client).each do |event|
next unless event.kind == Arcp::Job::EventKind::METRIC
next unless event.body.name == 'cost.budget.remaining'
puts "budget remaining: #{event.body.value} #{event.body.unit}"
end
begin
handle.get_result(client: client)
rescue Arcp::Errors::BudgetExhausted, Arcp::Errors::LeaseExpired => e
# Never retryable — resubmit with a fresh lease/budget instead.
warn "job ended: #{e.code} #{e.message}"
endAttach read-only to a job submitted elsewhere and observe its live stream (with optional history replay) without cancel authority.
Sync do
observer = Arcp::Client.open(
transport: dashboard_transport,
auth: { 'scheme' => 'bearer', 'token' => ENV.fetch('ARCP_TOKEN') },
client_name: 'dashboard'
)
running = observer.list_jobs(status: 'running', limit: 1).first
stream = observer.subscribe_job(job_id: running.job_id, from_event_seq: 0, history: true)
stream.each do |event|
puts "[#{running.job_id}] #{event.kind}"
end
observer.close
endCatch the typed error taxonomy and respect the retryable flag — LEASE_EXPIRED and BUDGET_EXHAUSTED are never retryable; a naive retry fails identically.
begin
handle = client.submit_job(agent: 'flaky', input: {})
handle.get_result(client: client)
rescue Arcp::Errors::LeaseExpired, Arcp::Errors::BudgetExhausted => e
raise e # resubmit with a fresh lease / budget instead
rescue Arcp::Error => e
if e.retryable?
# safe to retry with backoff (e.g. INTERNAL_ERROR, RATE_LIMITED, TIMEOUT)
retry_with_backoff(e)
else
raise
end
endARCP features this SDK negotiates during the hello/welcome handshake:
| Feature flag | Status |
|---|---|
heartbeat |
Supported |
ack |
Supported |
list_jobs |
Supported |
subscribe |
Supported |
lease_expires_at |
Supported |
cost.budget |
Supported |
model.use |
Supported |
provisioned_credentials |
Supported |
progress |
Supported |
result_chunk |
Supported |
agent_versions |
Supported |
ARCP is transport-agnostic. This SDK ships a WebSocket transport (default), a stdio transport for in-process child runtimes, and an in-memory transport for tests. WebSocket is the default for networked runtimes; stdio is used for in-process child runtimes. Select one by constructing the corresponding transport object and passing it to Arcp::Client.open(transport:, ...): Arcp::Transport::WebSocketTransport.new(connection: ws) for production (wrap an open Async::WebSocket::Connection, typically hosted under falcon), Arcp::Transport::StdioTransport for co-process agents, or Arcp::Transport::MemoryTransport.pair for in-process tests and embedded clients.
Full API reference — every type, method, and event payload — is in docs/ (YARD-generated reference under docs/api/, rebuilt with bundle exec rake docs).
This SDK speaks ARCP v1.1 (draft). The SDK follows semantic versioning independently of the protocol; the protocol version it negotiates is shown above and in session.hello. A runtime advertising a different ARCP MAJOR is not guaranteed compatible. Feature mismatches degrade gracefully: the effective feature set is the intersection of what the client and runtime advertise, and the SDK will not use a feature outside it.
See CONTRIBUTING.md. Protocol questions and proposed changes belong in the spec repository; SDK bugs and feature requests belong here.
Apache-2.0 — see LICENSE.