Expand description
Pillars: [AI][DevX]
Voice + LLM Interface for MockForge
This module provides voice input capability that allows users to build mocks conversationally using natural language commands. It leverages MockForge’s existing LLM infrastructure to interpret voice commands and generate mock APIs.
§Features
- Natural Language Command Parsing: Interpret voice commands using LLM
- OpenAPI Spec Generation: Generate OpenAPI 3.0 specs from voice commands
- Conversational Mode: Support multi-turn conversations for iterative refinement
- Single-Shot Mode: Process complete commands in one go
Issue #562 phase 7 moved the 6 leaf voice files into this crate. The 7th file,
workspace_builder.rs, stays in mockforge_core::voice_workspace because it
depends on multi_tenant, scenarios, workspace, contract_drift, and
reality_continuum — all still core-only. The mockforge_core::voice::*
shim consolidates both halves so callers see one unified public API.
§Do not extract voice_workspace into this crate
See the matching rationale in mockforge_core::voice (the shim) for the
full cycle diagram. Short version: voice_workspace calls the live
MultiTenantWorkspaceRegistry engine, and the 5 modules it depends on are
domain primitives that belong in core. Importing them here would invert
the established core → intelligence dep direction
(already in use via contract_drift re-exporting threat_modeling) and
reintroduce the cycle that Issue #562 phase 1 broke. Issue #562 is
complete; the location of voice_workspace is a feature, not debt.
Re-exports§
pub use command_parser::ApiRequirement;pub use command_parser::EntityRequirement;pub use command_parser::ParsedCommand;pub use command_parser::ParsedContinuumRule;pub use command_parser::ParsedDriftBudget;pub use command_parser::ParsedRealityContinuum;pub use command_parser::ParsedServiceBudget;pub use command_parser::ParsedWorkspaceCreation;pub use command_parser::ParsedWorkspaceScenario;pub use command_parser::PersonaRequirement;pub use command_parser::ScenarioRequirement;pub use command_parser::VoiceCommandParser;pub use conversation::ConversationContext;pub use conversation::ConversationManager;pub use conversation::ConversationState;pub use hook_transpiler::HookTranspiler;pub use spec_generator::VoiceSpecGenerator;pub use workspace_scenario_generator::GeneratedWorkspaceScenario;pub use workspace_scenario_generator::WorkspaceConfigSummary;pub use workspace_scenario_generator::WorkspaceScenarioGenerator;
Modules§
- command_
parser - LLM-based command parser for voice commands
- conversation
- Conversation state management for multi-turn voice interactions
- hook_
transpiler - Natural language to hook transpiler
- spec_
generator - OpenAPI spec generator from parsed voice commands
- workspace_
scenario_ generator - Workspace scenario generator