CoreRasurae
f2f7aa5dc9
Implement multiple Stage A/B/E/SEC milestones for the v3 lifecycle system: - Stage A5.3+A5.4: Add LifecycleActionModel and LifecyclePlanModel SQLAlchemy models with to_domain()/from_domain() conversion methods - Stage A5.6: Implement ActionRepository with full CRUD, namespace/state queries, referential integrity checks, and retry decorator - Stage E1: Add subplan domain models (ExecutionMode, SubplanMergeStrategy, SubplanConfig, SubplanStatus, SubplanAttempt, SubplanFailureHandler) with computed properties on Plan (is_subplan, is_root_plan, depth, has_subplans) - Stage A6: Add AutomationLevel enum (MANUAL, REVIEW_BEFORE_APPLY, FULL_AUTOMATION), settings integration, PlanLifecycleService auto-progression, pause/resume, and CLI commands (--automation-level, set-automation-level) - Stage SEC1: Remove eval()/exec() from stream_router.py, replace with named operation and transform registries; code blocks and unregistered transforms now raise StreamRoutingError - Add langchain-anthropic dependency - Update BDD tests for security changes and relax ADR directory requirement
77 lines
3.5 KiB
Gherkin
77 lines
3.5 KiB
Gherkin
Feature: Stream Router agent and tool coverage
|
|
As a developer
|
|
I want Behave coverage for SimpleToolAgent and SimpleLLMAgent
|
|
So that uncovered agent branches are exercised
|
|
|
|
Scenario: SimpleToolAgent returns content when tool code is missing
|
|
Given a SimpleToolAgent with a tool missing code
|
|
When I process "payload" through the SimpleToolAgent
|
|
Then the SimpleToolAgent result should be "payload"
|
|
|
|
Scenario: SimpleToolAgent rejects tool with code block
|
|
Given a SimpleToolAgent with a code block tool
|
|
When I process "payload" through the SimpleToolAgent expecting rejection
|
|
Then the SimpleToolAgent should raise a routing error about code blocks
|
|
|
|
Scenario: SimpleToolAgent rejects code block via process_message_sync
|
|
Given a SimpleToolAgent with a code block tool
|
|
When I call process_message_sync with "payload" expecting rejection
|
|
Then the SimpleToolAgent should raise a routing error about code blocks
|
|
|
|
# TODO: Uncomment when step definitions are implemented
|
|
# Scenario: SimpleLLMAgent returns empty prompt for blank template
|
|
# Given a SimpleLLMAgent for prompt rendering
|
|
# When I render the template "<empty>" with context:
|
|
# | key | value |
|
|
# | name | Ada |
|
|
# Then the rendered prompt should be "<empty>"
|
|
#
|
|
# TODO: Uncomment when step definitions are implemented
|
|
# Scenario: SimpleLLMAgent returns raw template when environment is missing
|
|
# Given a SimpleLLMAgent for prompt rendering
|
|
# And the SimpleLLMAgent template environment is "none"
|
|
# When I render the template "Hello" with context:
|
|
# | key | value |
|
|
# | name | Ada |
|
|
# Then the rendered prompt should be "Hello"
|
|
#
|
|
# TODO: Uncomment when step definitions are implemented
|
|
# Scenario: SimpleLLMAgent renders template when environment succeeds
|
|
# Given a SimpleLLMAgent for prompt rendering
|
|
# And the SimpleLLMAgent template environment is "working"
|
|
# When I render the template "Hello {{ name }}" with context:
|
|
# | key | value |
|
|
# | name | Ada |
|
|
# Then the rendered prompt should be "Hello Ada"
|
|
#
|
|
# TODO: Uncomment when step definitions are implemented
|
|
# Scenario: SimpleLLMAgent falls back to template on render failure
|
|
# Given a SimpleLLMAgent for prompt rendering
|
|
# And the SimpleLLMAgent template environment is "failing"
|
|
# When I render the template "Hello {{ name }}" with context:
|
|
# | key | value |
|
|
# | name | Ada |
|
|
# Then the rendered prompt should be "Hello {{ name }}"
|
|
|
|
Scenario: SimpleLLMAgent resolves LLM with configured kwargs and caching
|
|
Given a SimpleLLMAgent with provider config
|
|
And a stub provider registry is installed
|
|
When I resolve the LLM twice
|
|
Then the registry should be called once with the configured kwargs
|
|
And the resolved LLM should be cached
|
|
|
|
Scenario: SimpleLLMAgent raises when LangChain messages are unavailable
|
|
Given a SimpleLLMAgent with provider config
|
|
And a stub provider registry is installed
|
|
And LangChain message classes are unavailable
|
|
When I process "ping" through the LLM agent
|
|
Then I should receive a stream routing error about missing messages
|
|
|
|
Scenario: SimpleLLMAgent process_message_sync returns LLM content
|
|
Given a SimpleLLMAgent with system prompt and provider config
|
|
And a stub provider registry is installed
|
|
And LangChain message stubs are installed
|
|
When I process 123 through the LLM agent via process_message_sync
|
|
Then the LLM should receive system and human messages
|
|
And the LLM agent result should be "llm-output"
|