Module: Phronomy::Agent::Concerns::Suspendable Private

Included in:
Base
Defined in:
lib/phronomy/agent/concerns/suspendable.rb

Overview

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

Adds suspend/resume and tool-approval support to an agent.

Included in Base. When a tool decorated with +requires_approval true+ is called and no synchronous approval handler has been registered, the invocation is suspended and a Phronomy::Agent::Checkpoint is returned so the caller can resume later.

Instance Method Summary collapse

Instance Method Details

#checkpoint_store=(store) ⇒ void

This method returns an undefined value.

Sets the idempotency store used to guard against duplicate resumes.

The store must respond to:

Defaults to a per-instance Phronomy::Agent::CheckpointStore (in-memory, not thread-safe). Assign a shared persistent store when resuming across processes (e.g. Redis-backed). Custom stores are responsible for ensuring thread-safety if shared across threads.

Parameters:

  • store (#consumed?, #consume!) 


65
66
67
 
# File 'lib/phronomy/agent/concerns/suspendable.rb', line 65

def checkpoint_store=(store)
  @checkpoint_store = store
end

#on_approval_required(&block) ⇒ 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 a callback that is invoked before executing any tool that has +requires_approval true+ set. The block receives the tool name (String) and the arguments Hash, and must return a truthy value to allow execution. Returning a falsy value causes the tool to return a denial message instead of executing.

When no handler is registered and a tool with +requires_approval+ is called, #invoke returns a suspended result hash containing a Phronomy::Agent::Checkpoint. Call #resume to continue execution after obtaining an approval decision from the user or an external system.

Examples:

Synchronous handler

agent = MyAgent.new
agent.on_approval_required { |tool_name, args| prompt_user(tool_name, args) }

Returns:

  • (self) 


32
33
34
35
 
# File 'lib/phronomy/agent/concerns/suspendable.rb', line 32

def on_approval_required(&block)
  @approval_handler = block
  self
end

#resume(checkpoint, approved:, config: {}) ⇒ Hash

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.

Resumes a previously suspended invocation from a Phronomy::Agent::Checkpoint.

This method reconstructs the conversation state captured at suspension time, injects the tool result (executed or denied), and continues the LLM loop until it produces a final answer.

Parameters:

  • checkpoint (Phronomy::Agent::Checkpoint) —

    the checkpoint returned by the suspended #invoke call

  • approved (Boolean) —

    +true+ to execute the pending tool; +false+ to inject a denial message and let the LLM handle it gracefully

  • config (Hash) (defaults to: {}) —

    same runtime options as #invoke

Returns:

  • (Hash) —

    +{ output: String, suspended: false, messages: Array, usage: Phronomy::TokenUsage }+ or +{ output: nil, suspended: true, checkpoint: Phronomy::Agent::Checkpoint, messages: Array }+ when a second approval-required tool is encountered during continuation

Raises:



86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
 
# File 'lib/phronomy/agent/concerns/suspendable.rb', line 86

def resume(checkpoint, approved:, config: {})
  # Guard against duplicate resumes using the idempotency store.
  _checkpoint_store.consume!(checkpoint.checkpoint_id)
  # Build a fresh chat with all tools registered.
  chat = build_chat

  # Re-apply system instructions and register tools so the LLM has the
  # same persona/context as the original invocation. build_context
  # includes all tool classes (static + handoff) via add_capability.
  context = build_context(checkpoint.original_input, messages: [])
  apply_instructions(chat, context[:system]) if context[:system]
  (context[:tool_classes] || []).each { |tc| chat.with_tool(prepare_tool_class(tc)) }

  # Restore the full conversation (history + user + assistant with tool call).
  checkpoint.messages.each { |msg| chat.messages << msg }

  # Determine the tool result: execute it or inject a denial string.
  tool_result =
    if approved
      tool_instance = chat.tools[checkpoint.pending_tool_name.to_sym]
      tool_instance ? tool_instance.call(checkpoint.pending_tool_args) : "Tool not found."
    else
      "Tool execution denied."
    end

  # Inject the tool result so the LLM can continue.
  chat.add_message(
    role: :tool,
    content: tool_result.to_s,
    tool_call_id: checkpoint.pending_tool_call_id
  )

  # Re-register the suspension hook so that any further requires_approval
  # tools encountered during continuation are intercepted rather than
  # executed without approval (cascading / chained approval scenario).
  _register_suspension_hook!(chat)

  # Continue the LLM loop. Rescue SuspendSignal so that a second
  # approval-required tool produces a new checkpoint instead of running
  # without consent.
  begin
    response = chat.complete
  rescue SuspendSignal => signal
    new_checkpoint = Checkpoint.new(
      checkpoint_id: SecureRandom.uuid,
      agent_class: self.class.name,
      requested_at: Time.now.utc,
      thread_id: checkpoint.thread_id,
      original_input: checkpoint.original_input,
      messages: chat.messages.dup,
      pending_tool_name: signal.tool_name,
      pending_tool_args: signal.args,
      pending_tool_call_id: signal.tool_call_id
    )
    return {output: nil, suspended: true, checkpoint: new_checkpoint, messages: chat.messages}
  end

  output = response.content
  usage = Phronomy::TokenUsage.from_tokens(response.tokens)

  run_output_guardrails!(output)

  {output: output, suspended: false, messages: chat.messages, usage: usage}
end

#scope_policy=(policy) ⇒ void

This method returns an undefined value.

Registers a scope policy callable for this agent instance.

The callable receives +(tool_class, scope, agent)+ and must return +:allow+, +:reject+, or +:approve+.

Examples:

Reject all write-scoped tools

agent.scope_policy = ->(_tc, scope, _agent) { scope == :write ? :reject : :allow }

Parameters:

  • policy (#call) 


48
49
50
 
# File 'lib/phronomy/agent/concerns/suspendable.rb', line 48

def scope_policy=(policy)
  @scope_policy = policy
end