OptionalbackoffBackoff strategy for retrying failed operations.
Defines the logic for retrying failed operations when the agent encounters errors. This includes retrying function calls, API requests, and other operations that may fail due to temporary issues.
OptionallocaleLanguage and cultural locale preference for AI agent communication.
Configures the language and cultural context for all AI assistant responses throughout the vibe coding process. When specified, the agent will communicate in the preferred language while respecting cultural conventions and linguistic nuances during requirements gathering, progress updates, and guidance provision.
The locale setting also influences the agent's understanding of regional business practices, regulatory considerations, and cultural expectations that may impact software design decisions and implementation approaches.
Common formats follow BCP 47 language tags such as "en-US", "ko-KR", "ja-JP", "zh-CN", "de-DE", etc. Platform-specific detection methods:
navigator.languageprocess.env.LANG.split(".")[0]OptionalretryOptionalruntimeEnable a single post-runtime realize retrigger when endpoint failures are detected.
When true and the test-phase runtime reports at least one
endpointFailures entry, orchestrateTest will call
orchestrateRealizeRuntimeRetrigger once after the first runtime run. The
retrigger regenerates only the failing operations (using the per-endpoint
digest as regeneration context), re-runs the runtime with the new files,
and replaces the first-attempt summary with the second-attempt result in
the testComplete event.
Disabled by default. No budget loop — exactly one extra attempt.
OptionaltimeoutMaximum execution time limit for each LLM API response in milliseconds.
Controls the maximum duration allowed for a single provider response. When set to a numeric value, the agent will automatically abort a stalled LLM request after the specified time limit, preventing stuck provider calls from blocking a phase indefinitely.
Setting this value to null disables the timeout, allowing the agent to
wait indefinitely for each provider response. This unlimited mode should be
used with caution, particularly in production environments where resource
management is critical.
The timeout applies to each vendor request, not to the entire AutoBE pipeline or to local tool execution inside a conversation. A long analyze, database, interface, realize, or test phase can therefore continue as long as each individual AI response completes within the limit.
OptionaltimezoneGeographic timezone for temporal context and time-sensitive operations.
Provides timezone awareness for proper handling of time-related considerations during the vibe coding process. This includes scheduling references, temporal business logic requirements, timestamp handling in generated code, and time-based communications that need to be contextualized for the user's local time.
Timezone awareness ensures that generated applications properly handle time zones, that scheduling-related requirements are interpreted correctly, and that temporal references in documentation and code comments reflect the user's geographical context.
Format follows IANA Time Zone Database identifiers such as
"America/New_York", "Asia/Seoul", "Europe/London", "Pacific/Auckland", etc.
Platform detection: Intl.DateTimeFormat().resolvedOptions().timeZone
Interface defining behavioral configuration for AutoBeAgent localization and context.
This interface customizes the agent's communication style, language preferences, and geographical context to provide personalized vibe coding experiences. The configuration enables culturally appropriate interactions and regionally aware development decisions throughout the automated development pipeline.
Locale and timezone settings influence not only the language of communication but also contextual understanding of regulatory requirements, business practices, and temporal considerations that may affect requirements analysis, database design, API specifications, and implementation approaches.
Author
Samchon