Class: Phronomy::Agent::Base

Inherits:
Object
  • Object
show all
Includes:
Concerns::BeforeCompletion, Concerns::ErrorTranslation, Concerns::Guardrailable, Concerns::Retryable, Concerns::Suspendable, Runnable
Defined in:
lib/phronomy/agent/base.rb

Overview

Base class for all Phronomy agents.

Subclass this to create a conversational agent powered by an LLM. DSL class methods configure the model, instructions, tools, memory, and retry behaviour. Instance methods handle invocation.

Examples:

Minimal agent

class GreetingAgent < Phronomy::Agent::Base
  model "gpt-4o-mini"
  instructions "You are a friendly greeter."
end
result = GreetingAgent.new.invoke("Hello!")
puts result[:output]

Agent with tools

class ResearchAgent < Phronomy::Agent::Base
  model "gpt-4o"
  instructions "You are a research assistant."
  tools WebSearchTool, CalculatorTool
  max_iterations 15
end

Direct Known Subclasses

ReactAgent, MultiAgent::Orchestrator

Instance Attribute Summary

Attributes included from Concerns::BeforeCompletion

#before_completion

Class Method Summary collapse

Instance Method Summary collapse

Methods included from Concerns::Suspendable

#on_approval_required, #resume, #scope_policy=

Methods included from Concerns::BeforeCompletion

included

Methods included from Concerns::Guardrailable

#add_input_guardrail, #add_output_guardrail

Methods included from Concerns::Retryable

included

Methods included from Runnable

#batch, #trace

Class Method Details

.cache_instructions(enabled = nil) ⇒ Object

When enabled, attaches Anthropic prompt-cache markers to the system message so that the fixed instructions are served from cache on subsequent turns, reducing input-token costs.

Only has an effect when the agent also declares provider :anthropic. The cache_control field is provider-specific (the format differs between Anthropic direct, Bedrock, etc.), so the agent must explicitly declare its provider via the DSL rather than having it inferred from the model name.

Examples:

class MyAgent < Phronomy::Agent::Base
  provider :anthropic
  cache_instructions true
end


320
321
322
323
324
325
326
 
# File 'lib/phronomy/agent/base.rb', line 320

def cache_instructions(enabled = nil)
  if enabled.nil?
    @cache_instructions
  else
    @cache_instructions = enabled
  end
end

.context_overhead(val = nil) ⇒ Object

Tokens reserved for the system prompt + tool definitions overhead. Subtract this from the context window before computing the memory budget.

Examples:

class MyAgent < Phronomy::Agent::Base
  context_overhead 500
end


370
371
372
373
374
375
376
 
# File 'lib/phronomy/agent/base.rb', line 370

def context_overhead(val = nil)
  if val.nil?
    @context_overhead || 0
  else
    @context_overhead = val.to_i
  end
end

.context_window(val = nil) ⇒ Object

Overrides the context window size used for token budget calculations. When set, this value takes precedence over the RubyLLM model registry, which is useful for locally-hosted models (e.g. LM Studio) where the actually-loaded context length may differ from the catalogue value.

Examples:

class MyAgent < Phronomy::Agent::Base
  context_window 4096
end


354
355
356
357
358
359
360
 
# File 'lib/phronomy/agent/base.rb', line 354

def context_window(val = nil)
  if val.nil?
    @context_window
  else
    @context_window = val.to_i
  end
end

.instructions(text = nil) { ... } ⇒ String, ...

Sets or reads the system instructions for this agent. Accepts a String, a Context::Instruction::PromptTemplate, or a block (Proc). When used as a reader (no argument, no block), returns the stored value.

Examples:

String instructions

class MyAgent < Phronomy::Agent::Base
  instructions "You are a helpful assistant."
end

Block instructions

class MyAgent < Phronomy::Agent::Base
  instructions { |input| "Answer in #{input[:lang]}." }
end

Parameters:

Yields:

  • optionally provide instructions as a block

Returns:



77
78
79
80
81
82
83
84
 
# File 'lib/phronomy/agent/base.rb', line 77

def instructions(text = nil, &block)
  if text || block_given?
    @instructions = text || block
  else
    return @instructions if instance_variable_defined?(:@instructions)
    superclass.respond_to?(:instructions) ? superclass.instructions : nil
  end
end

.invoke_timeout(val = nil) ⇒ Numeric?

Sets or reads the per-invocation timeout (in seconds) for EventLoop-mode agent calls. When set, +invoke+ raises TimeoutError if the agent does not finish within the given number of seconds.

Has no effect when EventLoop mode is disabled (direct invoke path). Defaults to +nil+ (no timeout). Inherited by subclasses; the most-specific definition wins.

When the timeout fires, a Concurrency::CancellationScope is cancelled and its token is propagated to the FSM config so that in-flight LLM, tool, and RAG calls observe cancellation via their +cancellation_token:+ keyword argument. +Phronomy::TimeoutError+ is raised to the caller.

Examples:

class MyAgent < Phronomy::Agent::Base
  invoke_timeout 30
end

Parameters:

  • val (Numeric, nil) (defaults to: nil)

Returns:

  • (Numeric, nil) 


239
240
241
242
243
244
245
246
247
248
249
250
 
# File 'lib/phronomy/agent/base.rb', line 239

def invoke_timeout(val = nil)
  if val.nil?
    return @invoke_timeout if defined?(@invoke_timeout)
    superclass.respond_to?(:invoke_timeout) ? superclass.invoke_timeout : nil
  else
    unless val.is_a?(Numeric) && val > 0
      raise ArgumentError,
        "invoke_timeout must be a positive number, got #{val.inspect}"
    end
    @invoke_timeout = val
  end
end

.max_iterations(val = nil) ⇒ Integer

Sets or reads the maximum number of LLM call cycles for ReAct agents. Each tool call and follow-up counts as one iteration. Defaults to 10.

Examples:

class MyAgent < Phronomy::Agent::Base
  max_iterations 5
end

Parameters:

  • val (Integer, nil) (defaults to: nil)

Returns:

  • (Integer) 


184
185
186
187
188
189
190
 
# File 'lib/phronomy/agent/base.rb', line 184

def max_iterations(val = nil)
  if val
    @max_iterations = val
  else
    @max_iterations || 10
  end
end

.max_output_tokens(val = nil) ⇒ Object

Tokens to reserve for the model's output. When nil, the model's max_output_tokens from the registry is used.

Examples:

class MyAgent < Phronomy::Agent::Base
  max_output_tokens 4096
end


336
337
338
339
340
341
342
 
# File 'lib/phronomy/agent/base.rb', line 336

def max_output_tokens(val = nil)
  if val.nil?
    @max_output_tokens
  else
    @max_output_tokens = val.to_i
  end
end

.max_parallel_tools(val = nil) ⇒ Integer

Sets or reads the maximum number of tool calls executed concurrently when the LLM returns multiple tool calls in a single response (ParallelToolChat mode, active inside an AgentFSM IO thread).

Defaults to 10. Set to 1 to force sequential execution. Inherited by subclasses; the most-specific definition wins.

Examples:

class MyAgent < Phronomy::Agent::Base
  max_parallel_tools 4
end

Parameters:

  • val (Integer, nil) (defaults to: nil)

Returns:

  • (Integer) 


206
207
208
209
210
211
212
213
214
215
216
217
 
# File 'lib/phronomy/agent/base.rb', line 206

def max_parallel_tools(val = nil)
  if val.nil?
    @max_parallel_tools ||
      (superclass.respond_to?(:max_parallel_tools) ? superclass.max_parallel_tools : 10)
  else
    unless val.is_a?(Integer) && val >= 1
      raise ArgumentError,
        "max_parallel_tools must be a positive Integer (>= 1), got #{val.inspect}"
    end
    @max_parallel_tools = val
  end
end

.model(name = nil) ⇒ String?

Sets or reads the LLM model identifier for this agent. When called without an argument, returns the stored model or the global default from Phronomy.configuration.

Examples:

class MyAgent < Phronomy::Agent::Base
  model "gpt-4o"
end

Parameters:

  • name (String, nil) (defaults to: nil) —

    model identifier (e.g. "gpt-4o", "claude-3-5-sonnet")

Returns:

  • (String, nil) —

    the model name when used as a reader



53
54
55
56
57
58
59
 
# File 'lib/phronomy/agent/base.rb', line 53

def model(name = nil)
  if name
    @model = name
  else
    @model || Phronomy.configuration.default_model
  end
end

.provider(name = nil) ⇒ Symbol?

Sets or reads the LLM provider for this agent. Required when using a model not registered in RubyLLM's model registry (e.g. locally-hosted models via LM Studio or Ollama).

Examples:

class MyAgent < Phronomy::Agent::Base
  model "openai/gpt-oss-20b"
  provider :openai
end

Parameters:

  • name (Symbol, nil) (defaults to: nil) —

    e.g. +:openai+, +:anthropic+, +:ollama+

Returns:

  • (Symbol, nil) 


147
148
149
150
151
152
153
154
 
# File 'lib/phronomy/agent/base.rb', line 147

def provider(name = nil)
  if name
    @provider = name
  else
    return @provider if instance_variable_defined?(:@provider)
    superclass.respond_to?(:provider) ? superclass.provider : nil
  end
end

.static_knowledge(*sources) ⇒ Object

Registers one or more static knowledge sources on the agent class. Static source content is fetched and memoized at the class level the first time +invoke+ is called. The cache persists for the lifetime of the process; call static_knowledge_refresh! to force a reload.

Examples:

class PolicyAgent < Phronomy::Agent::Base
  static_knowledge Phronomy::Agent::Context::Knowledge::StaticKnowledge.new(POLICY_TEXT)
end

Parameters:



263
264
265
266
267
268
 
# File 'lib/phronomy/agent/base.rb', line 263

def static_knowledge(*sources)
  @static_knowledge_sources = sources.flatten
  # Invalidate the cached chunks so the new sources are fetched on
  # the next call to static_knowledge_chunks.
  @static_knowledge_chunks = nil
end

.static_knowledge_chunks ⇒ Array<Hash>

Returns the fetched content from all static knowledge sources. Results are cached at the class level so that each source is fetched only once regardless of how many times the agent is invoked.

Returns:

  • (Array<Hash>) 


282
283
284
285
286
 
# File 'lib/phronomy/agent/base.rb', line 282

def static_knowledge_chunks
  @static_knowledge_chunks ||= static_knowledge_sources.flat_map { |ks|
    ks.fetch(query: nil)
  }
end

.static_knowledge_refresh! ⇒ nil

Clears the class-level knowledge cache so that the next +invoke+ call re-fetches content from all registered static knowledge sources.

Call this method when the underlying knowledge source has been updated at runtime (e.g. a file was rewritten, a DB record changed) and you want the agent to pick up the new content without restarting the process.

Examples:

Refresh after updating a knowledge file

MyAgent.static_knowledge_refresh!

Returns:

  • (nil) 


300
301
302
 
# File 'lib/phronomy/agent/base.rb', line 300

def static_knowledge_refresh!
  @static_knowledge_chunks = nil
end

.static_knowledge_sources ⇒ Array<Phronomy::Agent::Context::Knowledge::Base>

Returns the registered static knowledge sources.

Returns:



273
274
275
 
# File 'lib/phronomy/agent/base.rb', line 273

def static_knowledge_sources
  @static_knowledge_sources || []
end

.temperature(val = nil) ⇒ Float?

Sets or reads the sampling temperature sent to the LLM. When nil, the provider's default is used.

Examples:

class MyAgent < Phronomy::Agent::Base
  temperature 0.2
end

Parameters:

  • val (Float, nil) (defaults to: nil) —

    temperature (0.0 to 2.0 depending on provider)

Returns:

  • (Float, nil) 


166
167
168
169
170
171
172
 
# File 'lib/phronomy/agent/base.rb', line 166

def temperature(val = nil)
  if val
    @temperature = val
  else
    @temperature
  end
end

.tool_aliases ⇒ Hash{Class => String}

Returns the alias map registered via the hash form of .tools. Merges parent class aliases so subclasses inherit their parent's mappings. Subclass-specific aliases take precedence over parent aliases.

Returns:

  • (Hash{Class => String}) 


126
127
128
129
130
131
132
133
 
# File 'lib/phronomy/agent/base.rb', line 126

def tool_aliases
  own = @tool_aliases || {}
  if superclass.respond_to?(:tool_aliases)
    superclass.tool_aliases.merge(own)
  else
    own
  end
end

.tools(*args) ⇒ Object

Registers tool classes for this agent.

Accepts either a splat of classes (backward-compatible) or a Hash mapping each class to an explicit alias name (String) or nil (use tool's own name). The alias form is useful when two tools share the same auto-generated name (e.g. two SearchTool classes from different modules).

Examples:

Splat form (no alias)

tools WeatherTool, TimeTool

Hash form (with optional per-tool alias)

tools(
  Weather::SearchTool => "weather_search",
  Places::SearchTool  => "places_search",
  CurrentTimeTool     => nil
)


103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
 
# File 'lib/phronomy/agent/base.rb', line 103

def tools(*args)
  if args.empty?
    if instance_variable_defined?(:@tools)
      return @tools
    end
    return superclass.respond_to?(:tools) ? superclass.tools : []
  end

  if args.length == 1 && args.first.is_a?(Hash)
    hash = args.first
    @tools = hash.keys
    @tool_aliases = hash.transform_values { |v| v&.to_s }.reject { |_, v| v.nil? }
  else
    @tools = args
    @tool_aliases = {}
  end
end

Instance Method Details

#_add_handoff_tool(tool_class) ⇒ self

This method is part of a private API. You should avoid using this method if possible, as it may be removed or be changed in the future.

Registers an anonymous handoff tool class on this agent instance. Called by Runner during construction when routes are configured.

Parameters:

Returns:

  • (self) 


384
385
386
387
388
 
# File 'lib/phronomy/agent/base.rb', line 384

def _add_handoff_tool(tool_class)
  @_handoff_tools ||= []
  @_handoff_tools << tool_class
  self
end

#_handoff_tools ⇒ Array<Class>

This method is part of a private API. You should avoid using this method if possible, as it may be removed or be changed in the future.

Returns handoff tool classes registered on this instance by Runner.

Returns:

  • (Array<Class>) 


393
394
395
 
# File 'lib/phronomy/agent/base.rb', line 393

def _handoff_tools
  @_handoff_tools || []
end

#context_version_cache ⇒ Object

This method is part of a private API. You should avoid using this method if possible, as it may be removed or be changed in the future.

Deprecated.

The context version cache has been removed. Returns nil. Retained for backward compatibility with callers using safe navigation (+&.reset+).



566
567
568
 
# File 'lib/phronomy/agent/base.rb', line 566

def context_version_cache
  nil
end

#invoke(input, messages: [], thread_id: nil, config: {}, invocation_context: nil) ⇒ Hash

Invokes the agent with the given input and returns a result Hash. Applies the retry policy configured via retry_policy when transient errors occur. GuardrailError is never retried.

Examples:

Normal invocation

result = MyAgent.new.invoke("What is Ruby?")
puts result[:output]

Multi-turn conversation

result1 = agent.invoke("Hi, I'm Alice.")
result2 = agent.invoke("What's my name?", messages: result1[:messages])

Suspend / resume flow

result = agent.invoke("Perform task X")
if result[:suspended]
  result = agent.resume(result[:checkpoint], approved: true)
end
puts result[:output]

With InvocationContext (deadline-based timeout)

ctx = Phronomy::InvocationContext.new(
  thread_id: "conv-123",
  deadline: Phronomy::Concurrency::Deadline.in(30),
  task_id: SecureRandom.uuid
)
result = MyAgent.new.invoke("Hello", invocation_context: ctx)

Parameters:

  • input (String, Hash) —

    the user message; a Hash may supply +:message+, +:query+, or +:user+ as the text key, plus any template variables consumed by the configured instructions template.

  • messages (Array<RubyLLM::Message>) (defaults to: []) —

    conversation history from a previous invocation. The application owns and persists this array; pass it on every turn to maintain multi-turn context.

  • thread_id (String, nil) (defaults to: nil) —

    conversation thread identifier, forwarded to the compaction context when on_compact is configured.

  • config (Hash) (defaults to: {}) —

    additional runtime options: +:user_id+ (+String+, optional) — caller identity forwarded to the tracer +:session_id+ (+String+, optional) — session identity forwarded to the tracer

  • invocation_context (Phronomy::InvocationContext, nil) (defaults to: nil) —

    optional first-class context object. When present, +thread_id+, +cancellation_token+, and +deadline+ are derived from it (existing +config:+ keys take precedence as backward-compat aliases). The object is also stored in +config[:invocation_context]+ so that +task_id+ / +parent_task_id+ appear in trace spans automatically.

Returns:

  • (Hash) —

    +{ output: String, messages: Array, usage: Phronomy::TokenUsage }+, or +{ output: nil, suspended: true, checkpoint: Phronomy::Agent::Checkpoint, messages: Array }+ when the invocation was suspended awaiting tool approval.

Raises:



441
442
443
444
445
446
447
448
449
450
451
 
# File 'lib/phronomy/agent/base.rb', line 441

def invoke(input, messages: [], thread_id: nil, config: {}, invocation_context: nil)
  if invocation_context
    thread_id, config = _apply_invocation_context(thread_id, config, invocation_context)
  end
  if Phronomy.configuration.event_loop
    _invoke_via_event_loop(input, messages: messages, thread_id: thread_id, config: config)
  else
    _check_scheduler_reentrancy
    invoke_async(input, messages: messages, thread_id: thread_id, config: config).await
  end
end

#invoke_async(input, messages: [], thread_id: nil, config: {}, invocation_context: nil) ⇒ Phronomy::Task

Invokes this agent asynchronously and returns a Task.

This is the primary async entry point. #invoke is a synchronous wrapper that calls this method and blocks the caller until the task completes. Calling #invoke from inside an active scheduler task raises SchedulerReentrancyError; use +invoke_async+ directly in that context.

The task is registered with the Runtime task registry so Runtime#shutdown drains in-flight invocations before process exit.

Examples:

task = agent.invoke_async("Hello!")
result = task.await   # => { output: "...", messages: [...], usage: ... }

Parameters:

  • input (String, Hash)
  • messages (Array) (defaults to: [])
  • thread_id (String, nil) (defaults to: nil)
  • config (Hash) (defaults to: {})
  • invocation_context (Phronomy::InvocationContext, nil) (defaults to: nil)

Returns:



475
476
477
478
479
480
481
482
483
484
485
486
487
488
 
# File 'lib/phronomy/agent/base.rb', line 475

def invoke_async(input, messages: [], thread_id: nil, config: {}, invocation_context: nil)
  if invocation_context
    thread_id, config = _apply_invocation_context(thread_id, config, invocation_context)
  end
  bp = Phronomy.configuration.backpressure
  on_full = (bp == :raise) ? :reject : (bp || :wait)
  bp_timeout = Phronomy.configuration.backpressure_timeout
  gate = Phronomy::Runtime.instance.gate(:agent)
  Phronomy::Runtime.instance.spawn(name: "agent-#{(self.class.name || "anonymous").downcase}-async") do
    gate.acquire(on_full: on_full, timeout: bp_timeout) do
      _invoke_impl(input, messages: messages, thread_id: thread_id, config: config)
    end
  end
end

#run_as_child(input, ctx:, messages: [], config: {}) ⇒ nil

Registers this agent as a child AgentFSM inside the given Workflow context.

Use this method from a Workflow entry action (running on the EventLoop thread) instead of #invoke, which would raise a deadlock error because +invoke+ blocks on a +Thread::Queue+ when EventLoop mode is active.

The agent runs asynchronously in a background IO thread. When it finishes, the parent FSMSession receives a +:child_completed+ event whose payload is the result hash +{ output:, messages:, usage: }+. Declare an +on: :child_completed+ transition in your Workflow to advance to the next state.

The result is delivered exclusively as the +:child_completed+ event payload. The parent Workflow task is the sole owner of the parent +WorkflowContext+ and applies the result after receiving the event — no background thread writes to the parent context directly.

Examples:

entry :run_agent, ->(ctx) { MyAgent.new.run_as_child(ctx.query, ctx: ctx) }
transition from: :run_agent, on: :child_completed, to: :process_result

Parameters:

  • input (String, Hash) —

    user input passed to the agent

  • ctx (Object) —

    a WorkflowContext that responds to +#thread_id+

  • messages (Array) (defaults to: []) —

    prior conversation history

  • config (Hash) (defaults to: {}) —

    invocation config (forwarded to +_invoke_impl+)

Returns:

  • (nil) —

    the caller must not wait on any return value; the result arrives as a +:child_completed+ event

Raises:



518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
 
# File 'lib/phronomy/agent/base.rb', line 518

def run_as_child(input, ctx:, messages: [], config: {})
  unless Phronomy.configuration.event_loop
    raise Phronomy::Error,
      "run_as_child requires EventLoop mode. " \
      "Enable with: Phronomy.configure { |c| c.event_loop = true }"
  end

  fsm = Agent::FSM.new(
    agent: self,
    input: input,
    messages: messages,
    thread_id: "#{ctx.thread_id}_agent_#{SecureRandom.uuid}",
    config: config,
    parent_id: ctx.thread_id
  )
  Phronomy::EventLoop.instance.enqueue_child(fsm)
  nil
end

#stream(input, messages: [], thread_id: nil, config: {}) {|Phronomy::Agent::StreamEvent| ... } ⇒ Hash

Streaming version of #invoke. Yields StreamEvent objects as they are produced by the underlying LLM.

Events emitted (in order): :token — each content delta from the LLM :tool_call — when the LLM requests a tool (ReactAgent subclasses only) :tool_result — after a tool completes (ReactAgent subclasses only) :done — final event carrying output, messages, and usage :error — if an unrecoverable error occurs

Parameters:

  • input (String, Hash) —

    same as #invoke

  • messages (Array<RubyLLM::Message>) (defaults to: []) —

    same as #invoke

  • thread_id (String, nil) (defaults to: nil) —

    same as #invoke

  • config (Hash) (defaults to: {}) —

    same as #invoke

Yields:

Returns:

  • (Hash) —

    { output:, messages:, usage: } — same as #invoke



554
555
556
557
558
559
560
561
 
# File 'lib/phronomy/agent/base.rb', line 554

def stream(input, messages: [], thread_id: nil, config: {}, &block)
  return invoke(input, messages: messages, thread_id: thread_id, config: config) unless block

  _stream_impl(input, messages: messages, thread_id: thread_id, config: config, &block)
rescue => e
  block&.call(StreamEvent.new(type: :error, payload: {error: e}))
  raise
end