rails_console_ai 0.26.0 → 0.28.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a2194416a93ce8522de376169eb627b90c8b2477ba253f0f1b877144251f9ee6
-  data.tar.gz: 65dcbeb6eef9dd2641181529aa671058e85a6cd1c5263acd51678c52c89d3567
+  metadata.gz: 50a2c9dce686cffa315e1fb5004f0ad0a9cbc67459d3a546b28ad1b95d0d1798
+  data.tar.gz: 7eea0529e3a3e4f9a4d60cf8e6b882e5dc825052c3793817013cf5cc6e617d22
 SHA512:
-  metadata.gz: e4d5a1f2fe8ef6ae4593829ac1a8997eeeac652ea0d0a491f91a64b8e45a246d024a10aa269e853949e0d106779d4f1b9d671af21057bd6af802ff6a541ac15d
-  data.tar.gz: ec0e0d25f7bb2605a7c4180ebc56ab56d8f590d3632c98310ec65c3aecd987d04118e7199534a8ea5a1a8023a505de9ac467ce05d01393df667c2436bcebc68c
+  metadata.gz: 4ed2a4ad47456f7e28682e0302d346fd04e3778433015855fc2199d4b1dc3077e8ec96efec5eb1b9ae049f6ed5f8c5ac6944e354488313b1f20e8f4354a0bdd1
+  data.tar.gz: c3a9d0faaada962952b9995ac0e07b79384536dc03970c0da17fe0e2eef843edea799d2dfa8698363a869d74e2aa7d86f2dbe80088234206693d86dd29ef45b4

data/CHANGELOG.md

@@ -2,6 +2,14 @@
 
 All notable changes to this project will be documented in this file.
 
+## [0.28.0]
+
+- Add `bin/smoke_model.rb` to smoke-test new models (plain, tool, parallel, cache checks)
+- Support Claude Opus 4.7 by omitting the `temperature` parameter for models that reject it
+- Show both estimated request tokens and total billed tokens in LLM round status
+- Auto-upgrade to thinking model on "think harder/deeper/carefully" phrases in Slack as well as console
+- Fix cancelled code execution state persisting into the next user turn
+
 ## [0.26.0]
 
 - Add sub-agent support

data/README.md

@@ -352,6 +352,39 @@ end
 
 Timeout is automatically raised to 300s minimum for local models to account for slower inference.
 
+### Testing a new model
+
+Before adopting a new Claude model, smoke-test it against the Anthropic or Bedrock provider with `bin/smoke_model.rb`. The script runs four checks and exits non-zero on any failure:
+
+| check    | what it verifies                                                                 |
+| -------- | -------------------------------------------------------------------------------- |
+| plain    | the model returns text for a basic prompt                                        |
+| tool     | a single tool call → tool result → final answer round-trip works                 |
+| parallel | the model issues multiple tool calls in one response when asked                  |
+| cache    | a long system prompt is written to and read from the prompt cache (with retry)  |
+
+```bash
+# Anthropic — provider inferred from the `claude-` prefix
+ANTHROPIC_API_KEY=sk-ant-... bin/smoke_model.rb --model claude-opus-4-7
+
+# Bedrock — provider inferred from the regional `us.anthropic.` prefix.
+# Requires the aws-sdk-bedrockruntime gem and AWS credentials in the environment.
+bin/smoke_model.rb --model us.anthropic.claude-opus-4-7
+
+# Bedrock in another region
+bin/smoke_model.rb --model eu.anthropic.claude-opus-4-7 --region eu-west-1
+
+# Subset of checks, e.g. when iterating on cache behavior
+bin/smoke_model.rb --model claude-sonnet-4-6 --checks cache
+
+# Force a provider when the model ID is ambiguous
+bin/smoke_model.rb --provider anthropic --model claude-opus-4-7
+```
+
+`DEBUG=1` enables the providers' raw request/response logging.
+
+If the model rejects a parameter the gem sends by default (e.g. opus-4-7 deprecated `temperature`), add the model ID to `Configuration::MODELS_WITHOUT_TEMPERATURE` in `lib/rails_console_ai/configuration.rb` so the providers omit the field.
+
 ## Configuration
 
 ```ruby

data/lib/rails_console_ai/channel/console.rb

@@ -207,6 +207,7 @@ module RailsConsoleAi
           break if input.nil?
 
           input = input.strip
+          input = input.force_encoding('UTF-8') if input.encoding == Encoding::ASCII_8BIT
           break if input.downcase == 'exit' || input.downcase == 'quit'
           next if input.empty?
 
@@ -222,8 +223,7 @@ module RailsConsoleAi
           # Add to Readline history
           Readline::HISTORY.push(input) unless input == Readline::HISTORY.to_a.last
 
-          # Auto-upgrade to thinking model on "think harder" phrases
-          @engine.upgrade_to_thinking_model if input =~ /think\s*harder/i
+          @engine.maybe_auto_upgrade_thinking(input)
 
           @engine.set_interactive_query(input)
           @engine.add_user_message(input)

data/lib/rails_console_ai/configuration.rb

@@ -1,3 +1,5 @@
+require 'set'
+
 module RailsConsoleAi
   class Configuration
     PROVIDERS = %i[anthropic openai local bedrock].freeze
@@ -18,6 +20,17 @@ module RailsConsoleAi
       'claude-opus-4-6'   => 4_096,
     }.freeze
 
+    # Models that reject the `temperature` parameter. Configuration#resolved_temperature
+    # returns nil for these so providers can omit the field from the request.
+    MODELS_WITHOUT_TEMPERATURE = Set.new(%w[
+      claude-opus-4-7
+      anthropic.claude-opus-4-7
+      us.anthropic.claude-opus-4-7
+      eu.anthropic.claude-opus-4-7
+      jp.anthropic.claude-opus-4-7
+      global.anthropic.claude-opus-4-7
+    ]).freeze
+
     attr_accessor :provider, :api_key, :model, :thinking_model, :max_tokens,
                   :auto_execute, :temperature,
                   :timeout, :debug, :max_tool_rounds,
@@ -179,6 +192,13 @@ module RailsConsoleAi
       DEFAULT_MAX_TOKENS.fetch(resolved_model, 4096)
     end
 
+    # Returns nil for models that reject the `temperature` parameter (e.g. opus-4-7).
+    # Providers should use this in place of @temperature.
+    def resolved_temperature
+      return nil if MODELS_WITHOUT_TEMPERATURE.include?(resolved_model)
+      @temperature
+    end
+
     def resolved_thinking_model
       return @thinking_model if @thinking_model && !@thinking_model.empty?
 

data/lib/rails_console_ai/conversation_engine.rb

@@ -110,6 +110,7 @@ module RailsConsoleAi
       init_interactive unless @interactive_start
       @channel.log_input(text) if @channel.respond_to?(:log_input)
       @interactive_query ||= text
+      maybe_auto_upgrade_thinking(text)
       @history << { role: :user, content: text }
 
       status = send_and_execute
@@ -249,7 +250,7 @@ module RailsConsoleAi
       output_id = @executor.store_output(result_str)
       if result_str.length > LARGE_OUTPUT_THRESHOLD
         preview = result_str[0, LARGE_OUTPUT_PREVIEW_CHARS]
-        context_msg += "\n#{preview}\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{result_str.length} chars — use recall_output tool with id #{output_id} to retrieve the full output]"
+        context_msg += "\n#{preview}\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{result_str.length} chars — use explore_output with output_id=#{output_id} for focused queries, or recall_output to expand in place]"
       elsif !output_parts.empty?
         context_msg += "\n#{result_str}"
       end
@@ -332,7 +333,7 @@ module RailsConsoleAi
             context_msg = "Code was executed (safety override). "
             if result_str.length > LARGE_OUTPUT_THRESHOLD
               context_msg += result_str[0, LARGE_OUTPUT_PREVIEW_CHARS]
-              context_msg += "\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{result_str.length} chars — use recall_output tool with id #{output_id} to retrieve the full output]"
+              context_msg += "\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{result_str.length} chars — use explore_output with output_id=#{output_id} for focused queries, or recall_output to expand in place]"
             else
               context_msg += result_str
             end
@@ -360,7 +361,7 @@ module RailsConsoleAi
           context_msg = "Code was executed. "
           if result_str.length > LARGE_OUTPUT_THRESHOLD
             context_msg += result_str[0, LARGE_OUTPUT_PREVIEW_CHARS]
-            context_msg += "\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{result_str.length} chars — use recall_output tool with id #{output_id} to retrieve the full output]"
+            context_msg += "\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{result_str.length} chars — use explore_output with output_id=#{output_id} for focused queries, or recall_output to expand in place]"
           else
             context_msg += result_str
           end
@@ -450,6 +451,13 @@ module RailsConsoleAi
       parts.compact.join("\n\n")
     end
 
+    AUTO_THINK_PATTERN = /\bthink\s+(harder|deeper|hard|carefully|more\s+carefully)\b/i
+
+    def maybe_auto_upgrade_thinking(text)
+      return unless text.is_a?(String) && text =~ AUTO_THINK_PATTERN
+      upgrade_to_thinking_model
+    end
+
     def upgrade_to_thinking_model
       config = RailsConsoleAi.configuration
       current = effective_model
@@ -777,6 +785,7 @@ module RailsConsoleAi
       require 'rails_console_ai/tools/registry'
       tools = tools_override || Tools::Registry.new(executor: @executor, channel: @channel)
       active_system_prompt = system_prompt || context
+      @executor.reset_cancelled! if @executor
       max_rounds = RailsConsoleAi.configuration.max_tool_rounds
       total_input = 0
       total_output = 0
@@ -796,19 +805,21 @@ module RailsConsoleAi
 
         if round == 0
           @channel.display_status("  Thinking...")
-        else
-          if last_thinking
-            last_thinking.split("\n").each do |line|
-              @channel.display_thinking("  #{line}")
-            end
+        elsif last_thinking
+          last_thinking.split("\n").each do |line|
+            @channel.display_thinking("  #{line}")
           end
-          @channel.display_status("  #{llm_status(round, messages, total_input, last_thinking, last_tool_names)}")
         end
 
         # Trim large tool outputs between rounds to prevent context explosion.
         # The LLM can still retrieve omitted outputs via recall_output.
         messages = trim_large_outputs(messages) if round > 0
 
+        if round > 0
+          req_tokens = estimate_request_tokens(messages)
+          @channel.display_status("  #{llm_status(round, messages, req_tokens, total_input, last_thinking, last_tool_names)}")
+        end
+
         if RailsConsoleAi.configuration.debug
           debug_pre_call(round, messages, active_system_prompt, tools, total_input, total_output)
         end
@@ -903,7 +914,7 @@ module RailsConsoleAi
           tool_msg[:output_id] = output_id
           if full_text.length > LARGE_OUTPUT_THRESHOLD
             truncated = full_text[0, LARGE_OUTPUT_PREVIEW_CHARS]
-            truncated += "\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{full_text.length} chars — use recall_output tool with id #{output_id} to retrieve the full output]"
+            truncated += "\n\n[Output truncated at #{LARGE_OUTPUT_PREVIEW_CHARS} of #{full_text.length} chars — use explore_output with output_id=#{output_id} for focused queries, or recall_output to expand in place]"
             tool_msg = provider.format_tool_result(tc[:id], truncated)
             tool_msg[:output_id] = output_id
           end
@@ -1012,6 +1023,11 @@ module RailsConsoleAi
 
     # --- Formatting helpers ---
 
+    def estimate_request_tokens(messages)
+      chars = messages.sum { |m| (m[:content] || m['content']).to_s.length }
+      chars / 4
+    end
+
     def format_tokens(count)
       if count >= 1_000_000
         "#{(count / 1_000_000.0).round(1)}M"
@@ -1041,6 +1057,10 @@ module RailsConsoleAi
       when 'save_skill'     then "(\"#{args['name']}\")"
       when 'delete_skill'   then "(\"#{args['name']}\")"
       when 'recall_output'   then "(#{args['id']})"
+      when 'explore_output'
+        task_preview = args['task'].to_s[0, 80]
+        task_preview += '...' if args['task'].to_s.length > 80
+        "(id: #{args['output_id']}, \"#{task_preview}\")"
       when 'execute_plan'
         steps = args['steps']
         steps ? "(#{steps.length} steps)" : ''
@@ -1132,9 +1152,10 @@ module RailsConsoleAi
       str.length > max ? str[0..max] + '...' : str
     end
 
-    def llm_status(round, messages, tokens_so_far, last_thinking = nil, last_tool_names = [])
+    def llm_status(round, messages, req_tokens, total_billed, last_thinking = nil, last_tool_names = [])
       status = "Calling LLM (round #{round + 1}, #{messages.length} msgs"
-      status += ", ~#{format_tokens(tokens_so_far)} ctx" if tokens_so_far > 0
+      status += ", ~#{format_tokens(req_tokens)} ctx" if req_tokens > 0
+      status += ", ~#{format_tokens(total_billed)} total" if total_billed > 0
       status += ")"
       if !last_thinking && last_tool_names.any?
         counts = last_tool_names.tally
@@ -1409,7 +1430,7 @@ module RailsConsoleAi
     end
 
     def trim_message(msg)
-      ref = "[Output omitted — use recall_output tool with id #{msg[:output_id]} to retrieve]"
+      ref = "[Output omitted — use explore_output with output_id=#{msg[:output_id]} for focused queries, or recall_output to expand in place]"
 
       if msg[:content].is_a?(Array)
         trimmed_content = msg[:content].map do |block|

data/lib/rails_console_ai/executor.rb

@@ -206,6 +206,10 @@ module RailsConsoleAi
       @last_cancelled
     end
 
+    def reset_cancelled!
+      @last_cancelled = false
+    end
+
     def confirm_and_execute(code)
       return nil if code.nil? || code.strip.empty?
 

data/lib/rails_console_ai/providers/anthropic.rb

@@ -51,9 +51,10 @@ module RailsConsoleAi
         body = {
           model: config.resolved_model,
           max_tokens: config.resolved_max_tokens,
-          temperature: config.temperature,
           messages: format_messages(messages)
         }
+        temp = config.resolved_temperature
+        body[:temperature] = temp unless temp.nil?
         if system_prompt
           body[:system] = [
             { 'type' => 'text', 'text' => system_prompt, 'cache_control' => { 'type' => 'ephemeral' } }

data/lib/rails_console_ai/providers/bedrock.rb

@@ -41,13 +41,13 @@ module RailsConsoleAi
       private
 
       def call_api(messages, system_prompt: nil, tools: nil)
+        inference = { max_tokens: config.resolved_max_tokens }
+        temp = config.resolved_temperature
+        inference[:temperature] = temp unless temp.nil?
         params = {
           model_id: config.resolved_model,
           messages: format_messages(messages),
-          inference_config: {
-            max_tokens: config.resolved_max_tokens,
-            temperature: config.temperature
-          }
+          inference_config: inference
         }
         if system_prompt
           sys_blocks = [{ text: system_prompt }]

data/lib/rails_console_ai/providers/local.rb

@@ -21,9 +21,10 @@ module RailsConsoleAi
         body = {
           model: config.resolved_model,
           max_tokens: config.resolved_max_tokens,
-          temperature: config.temperature,
           messages: formatted
         }
+        temp = config.resolved_temperature
+        body[:temperature] = temp unless temp.nil?
         body[:tools] = tools.to_openai_format if tools
 
         estimated_input_tokens = estimate_tokens(formatted, system_prompt, tools)

data/lib/rails_console_ai/providers/openai.rb

@@ -51,9 +51,10 @@ module RailsConsoleAi
         body = {
           model: config.resolved_model,
           max_tokens: config.resolved_max_tokens,
-          temperature: config.temperature,
           messages: formatted
         }
+        temp = config.resolved_temperature
+        body[:temperature] = temp unless temp.nil?
         body[:tools] = tools.to_openai_format if tools
 
         json_body = JSON.generate(body)

data/lib/rails_console_ai/sub_agent.rb

@@ -12,12 +12,15 @@ module RailsConsoleAi
 
     attr_reader :input_tokens, :output_tokens, :model_used
 
-    def initialize(task:, agent_config:, binding_context:, parent_channel:, executor:)
+    def initialize(task:, agent_config:, binding_context:, parent_channel:, executor:,
+                   output_payload: nil, output_local_name: :output)
       @task = task
       @agent_config = agent_config || {}
       @binding_context = binding_context
       @parent_channel = parent_channel
       @parent_executor = executor
+      @output_payload = output_payload
+      @output_local_name = output_local_name
       @input_tokens = 0
       @output_tokens = 0
       @model_used = nil
@@ -29,7 +32,16 @@ module RailsConsoleAi
         task_label: @agent_config['name']
       )
 
-      executor = Executor.new(@binding_context, channel: channel)
+      effective_binding =
+        if @output_payload
+          b = @binding_context.eval("proc { binding }.call")
+          b.local_variable_set(@output_local_name, @output_payload)
+          b
+        else
+          @binding_context
+        end
+
+      executor = Executor.new(effective_binding, channel: channel)
       allowed_tools = @agent_config['tools'] ? Array(@agent_config['tools']) : nil
       tools = Tools::Registry.new(executor: executor, mode: :sub_agent, channel: channel, allowed_tools: allowed_tools)
       provider = build_provider

data/lib/rails_console_ai/tools/registry.rb

@@ -6,7 +6,7 @@ module RailsConsoleAi
       attr_reader :definitions, :last_sub_agent_usage
 
       # Tools that should never be cached (side effects or user interaction)
-      NO_CACHE = %w[ask_user save_memory delete_memory recall_memory execute_code execute_plan activate_skill save_skill delete_skill delegate_task].freeze
+      NO_CACHE = %w[ask_user save_memory delete_memory recall_memory execute_code execute_plan activate_skill save_skill delete_skill delegate_task explore_output].freeze
 
       def initialize(executor: nil, mode: :default, channel: nil, allowed_tools: nil)
         @executor = executor
@@ -188,7 +188,7 @@ module RailsConsoleAi
         if @executor
           register(
             name: 'recall_output',
-            description: 'Retrieve a previous code execution output that was omitted or truncated. The output will be expanded in place in the conversation. Use the output id shown in the "[Output omitted]" or "[Output truncated]" placeholder.',
+            description: 'Expand a previously omitted/truncated output back into this conversation\'s context, where it will persist for the rest of the session. Prefer `explore_output` if you only need a specific answer about the output — that keeps this conversation lean. Use `recall_output` only when you need the full content alongside other context here. Use the output id shown in the "[Output omitted]" or "[Output truncated]" placeholder.',
             parameters: {
               'type' => 'object',
               'properties' => {
@@ -204,7 +204,7 @@ module RailsConsoleAi
 
           register(
             name: 'recall_outputs',
-            description: 'Retrieve multiple previous code execution outputs that were omitted from the conversation. Use the output ids shown in "[Output omitted]" or "[Output truncated]" placeholders.',
+            description: 'Expand multiple previously omitted outputs back into this conversation. Prefer `explore_output` per-id for focused queries. Use the output ids shown in "[Output omitted]" or "[Output truncated]" placeholders.',
             parameters: {
               'type' => 'object',
               'properties' => {
@@ -214,6 +214,22 @@ module RailsConsoleAi
             },
             handler: ->(args) { "recall_outputs handled by conversation engine" }
           )
+
+          if @mode != :sub_agent
+            register(
+              name: 'explore_output',
+              description: 'Prefer this over recall_output when you have a specific question about a large omitted/truncated output (e.g. "find the item where X", "how many match Y", "what is the value at index N", "parse the JSON and return field Z"). Spawns a sub-agent with the full output bound to the local Ruby variable `output` (a String); the sub-agent runs execute_code against it and returns a concise answer. The full output does NOT enter this conversation.',
+              parameters: {
+                'type' => 'object',
+                'properties' => {
+                  'output_id' => { 'type' => 'integer', 'description' => 'The output id shown in the "[Output omitted]" or "[Output truncated]" placeholder.' },
+                  'task' => { 'type' => 'string', 'description' => 'The specific question or task. Be concrete — the sub-agent only sees this task and the output.' }
+                },
+                'required' => ['output_id', 'task']
+              },
+              handler: ->(args) { explore_output(args['output_id'].to_i, args['task']) }
+            )
+          end
         end
 
         unless @mode == :init
@@ -317,6 +333,45 @@ module RailsConsoleAi
         )
       end
 
+      EXPLORE_OUTPUT_AGENT_CONFIG = {
+        'name' => 'output-explorer',
+        'tools' => ['execute_code'],
+        'max_rounds' => 8,
+        'body' => <<~PROMPT.freeze
+          You are exploring a single chunk of captured tool output on behalf of the main assistant.
+
+          The full output is bound to the local variable `output` (a String). You do NOT see it
+          directly — it lives in Ruby memory. Use `execute_code` with Ruby to query it:
+            - `output.length`, `output.lines.count`
+            - `output[start, len]`, `output.lines[n]`
+            - `output.scan(/pattern/)`, `output.include?("...")`
+            - `JSON.parse(output)` if it looks like JSON, then drill in
+            - any other Ruby string/collection methods
+
+          Print only the specific slice or summary the task requires — never dump the whole `output`.
+          Return a concise factual answer. No preamble.
+        PROMPT
+      }.freeze
+
+      def explore_output(output_id, task)
+        require 'rails_console_ai/sub_agent'
+
+        payload = @executor.recall_output(output_id)
+        return "No output found with id #{output_id}" unless payload
+
+        sub = SubAgent.new(
+          task: task,
+          agent_config: EXPLORE_OUTPUT_AGENT_CONFIG,
+          binding_context: @executor.binding_context,
+          parent_channel: @channel,
+          executor: @executor,
+          output_payload: payload.dup
+        )
+        result = sub.run
+        @last_sub_agent_usage = { input: sub.input_tokens, output: sub.output_tokens, model: sub.model_used }
+        "Exploration result (#{sub.input_tokens + sub.output_tokens} tokens used, #{payload.length} chars explored):\n#{result}"
+      end
+
       def delegate_task(task, agent_name = nil)
         require 'rails_console_ai/sub_agent'
         require 'rails_console_ai/agent_loader'

data/lib/rails_console_ai/version.rb

@@ -1,3 +1,3 @@
 module RailsConsoleAi
-  VERSION = '0.26.0'.freeze
+  VERSION = '0.28.0'.freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rails_console_ai
 version: !ruby/object:Gem::Version
-  version: 0.26.0
+  version: 0.28.0
 platform: ruby
 authors:
 - Cortfr