feat: Phase 5.3 - Multi-Agent Learning Infrastructure
Implement intelligent agent learning from Knowledge Graph execution history
with per-task-type expertise tracking, recency bias, and learning curves.
## Phase 5.3 Implementation
### Learning Infrastructure (✅ Complete)
- LearningProfileService with per-task-type expertise metrics
- TaskTypeExpertise model tracking success_rate, confidence, learning curves
- Recency bias weighting: recent 7 days weighted 3x higher (exponential decay)
- Confidence scoring prevents overfitting: min(1.0, executions / 20)
- Learning curves computed from daily execution windows
### Agent Scoring Service (✅ Complete)
- Unified AgentScore combining SwarmCoordinator + learning profiles
- Scoring formula: 0.3*base + 0.5*expertise + 0.2*confidence
- Rank agents by combined score for intelligent assignment
- Support for recency-biased scoring (recent_success_rate)
- Methods: rank_agents, select_best, rank_agents_with_recency
### KG Integration (✅ Complete)
- KGPersistence::get_executions_for_task_type() - query by agent + task type
- KGPersistence::get_agent_executions() - all executions for agent
- Coordinator::load_learning_profile_from_kg() - core KG→Learning integration
- Coordinator::load_all_learning_profiles() - batch load for multiple agents
- Convert PersistedExecution → ExecutionData for learning calculations
### Agent Assignment Integration (✅ Complete)
- AgentCoordinator uses learning profiles for task assignment
- extract_task_type() infers task type from title/description
- assign_task() scores candidates using AgentScoringService
- Fallback to load-based selection if no learning data available
- Learning profiles stored in coordinator.learning_profiles RwLock
### Profile Adapter Enhancements (✅ Complete)
- create_learning_profile() - initialize empty profiles
- add_task_type_expertise() - set task-type expertise
- update_profile_with_learning() - update swarm profiles from learning
## Files Modified
### vapora-knowledge-graph/src/persistence.rs (+30 lines)
- get_executions_for_task_type(agent_id, task_type, limit)
- get_agent_executions(agent_id, limit)
### vapora-agents/src/coordinator.rs (+100 lines)
- load_learning_profile_from_kg() - core KG integration method
- load_all_learning_profiles() - batch loading for agents
- assign_task() already uses learning-based scoring via AgentScoringService
### Existing Complete Implementation
- vapora-knowledge-graph/src/learning.rs - calculation functions
- vapora-agents/src/learning_profile.rs - data structures and expertise
- vapora-agents/src/scoring.rs - unified scoring service
- vapora-agents/src/profile_adapter.rs - adapter methods
## Tests Passing
- learning_profile: 7 tests ✅
- scoring: 5 tests ✅
- profile_adapter: 6 tests ✅
- coordinator: learning-specific tests ✅
## Data Flow
1. Task arrives → AgentCoordinator::assign_task()
2. Extract task_type from description
3. Query KG for task-type executions (load_learning_profile_from_kg)
4. Calculate expertise with recency bias
5. Score candidates (SwarmCoordinator + learning)
6. Assign to top-scored agent
7. Execution result → KG → Update learning profiles
## Key Design Decisions
✅ Recency bias: 7-day half-life with 3x weight for recent performance
✅ Confidence scoring: min(1.0, total_executions / 20) prevents overfitting
✅ Hierarchical scoring: 30% base load, 50% expertise, 20% confidence
✅ KG query limit: 100 recent executions per task-type for performance
✅ Async loading: load_learning_profile_from_kg supports concurrent loads
## Next: Phase 5.4 - Cost Optimization
Ready to implement budget enforcement and cost-aware provider selection.
2026-01-11 13:03:53 +00:00
|
|
|
// Tasks API endpoints
|
|
|
|
|
|
2026-01-11 21:32:56 +00:00
|
|
|
use crate::api::ApiResult;
|
feat: Phase 5.3 - Multi-Agent Learning Infrastructure
Implement intelligent agent learning from Knowledge Graph execution history
with per-task-type expertise tracking, recency bias, and learning curves.
## Phase 5.3 Implementation
### Learning Infrastructure (✅ Complete)
- LearningProfileService with per-task-type expertise metrics
- TaskTypeExpertise model tracking success_rate, confidence, learning curves
- Recency bias weighting: recent 7 days weighted 3x higher (exponential decay)
- Confidence scoring prevents overfitting: min(1.0, executions / 20)
- Learning curves computed from daily execution windows
### Agent Scoring Service (✅ Complete)
- Unified AgentScore combining SwarmCoordinator + learning profiles
- Scoring formula: 0.3*base + 0.5*expertise + 0.2*confidence
- Rank agents by combined score for intelligent assignment
- Support for recency-biased scoring (recent_success_rate)
- Methods: rank_agents, select_best, rank_agents_with_recency
### KG Integration (✅ Complete)
- KGPersistence::get_executions_for_task_type() - query by agent + task type
- KGPersistence::get_agent_executions() - all executions for agent
- Coordinator::load_learning_profile_from_kg() - core KG→Learning integration
- Coordinator::load_all_learning_profiles() - batch load for multiple agents
- Convert PersistedExecution → ExecutionData for learning calculations
### Agent Assignment Integration (✅ Complete)
- AgentCoordinator uses learning profiles for task assignment
- extract_task_type() infers task type from title/description
- assign_task() scores candidates using AgentScoringService
- Fallback to load-based selection if no learning data available
- Learning profiles stored in coordinator.learning_profiles RwLock
### Profile Adapter Enhancements (✅ Complete)
- create_learning_profile() - initialize empty profiles
- add_task_type_expertise() - set task-type expertise
- update_profile_with_learning() - update swarm profiles from learning
## Files Modified
### vapora-knowledge-graph/src/persistence.rs (+30 lines)
- get_executions_for_task_type(agent_id, task_type, limit)
- get_agent_executions(agent_id, limit)
### vapora-agents/src/coordinator.rs (+100 lines)
- load_learning_profile_from_kg() - core KG integration method
- load_all_learning_profiles() - batch loading for agents
- assign_task() already uses learning-based scoring via AgentScoringService
### Existing Complete Implementation
- vapora-knowledge-graph/src/learning.rs - calculation functions
- vapora-agents/src/learning_profile.rs - data structures and expertise
- vapora-agents/src/scoring.rs - unified scoring service
- vapora-agents/src/profile_adapter.rs - adapter methods
## Tests Passing
- learning_profile: 7 tests ✅
- scoring: 5 tests ✅
- profile_adapter: 6 tests ✅
- coordinator: learning-specific tests ✅
## Data Flow
1. Task arrives → AgentCoordinator::assign_task()
2. Extract task_type from description
3. Query KG for task-type executions (load_learning_profile_from_kg)
4. Calculate expertise with recency bias
5. Score candidates (SwarmCoordinator + learning)
6. Assign to top-scored agent
7. Execution result → KG → Update learning profiles
## Key Design Decisions
✅ Recency bias: 7-day half-life with 3x weight for recent performance
✅ Confidence scoring: min(1.0, total_executions / 20) prevents overfitting
✅ Hierarchical scoring: 30% base load, 50% expertise, 20% confidence
✅ KG query limit: 100 recent executions per task-type for performance
✅ Async loading: load_learning_profile_from_kg supports concurrent loads
## Next: Phase 5.4 - Cost Optimization
Ready to implement budget enforcement and cost-aware provider selection.
2026-01-11 13:03:53 +00:00
|
|
|
use axum::{
|
|
|
|
|
extract::{Path, Query, State},
|
|
|
|
|
http::StatusCode,
|
|
|
|
|
response::IntoResponse,
|
|
|
|
|
Json,
|
|
|
|
|
};
|
|
|
|
|
use serde::Deserialize;
|
2026-01-11 21:32:56 +00:00
|
|
|
use vapora_shared::models::{Task, TaskPriority, TaskStatus};
|
feat: Phase 5.3 - Multi-Agent Learning Infrastructure
Implement intelligent agent learning from Knowledge Graph execution history
with per-task-type expertise tracking, recency bias, and learning curves.
## Phase 5.3 Implementation
### Learning Infrastructure (✅ Complete)
- LearningProfileService with per-task-type expertise metrics
- TaskTypeExpertise model tracking success_rate, confidence, learning curves
- Recency bias weighting: recent 7 days weighted 3x higher (exponential decay)
- Confidence scoring prevents overfitting: min(1.0, executions / 20)
- Learning curves computed from daily execution windows
### Agent Scoring Service (✅ Complete)
- Unified AgentScore combining SwarmCoordinator + learning profiles
- Scoring formula: 0.3*base + 0.5*expertise + 0.2*confidence
- Rank agents by combined score for intelligent assignment
- Support for recency-biased scoring (recent_success_rate)
- Methods: rank_agents, select_best, rank_agents_with_recency
### KG Integration (✅ Complete)
- KGPersistence::get_executions_for_task_type() - query by agent + task type
- KGPersistence::get_agent_executions() - all executions for agent
- Coordinator::load_learning_profile_from_kg() - core KG→Learning integration
- Coordinator::load_all_learning_profiles() - batch load for multiple agents
- Convert PersistedExecution → ExecutionData for learning calculations
### Agent Assignment Integration (✅ Complete)
- AgentCoordinator uses learning profiles for task assignment
- extract_task_type() infers task type from title/description
- assign_task() scores candidates using AgentScoringService
- Fallback to load-based selection if no learning data available
- Learning profiles stored in coordinator.learning_profiles RwLock
### Profile Adapter Enhancements (✅ Complete)
- create_learning_profile() - initialize empty profiles
- add_task_type_expertise() - set task-type expertise
- update_profile_with_learning() - update swarm profiles from learning
## Files Modified
### vapora-knowledge-graph/src/persistence.rs (+30 lines)
- get_executions_for_task_type(agent_id, task_type, limit)
- get_agent_executions(agent_id, limit)
### vapora-agents/src/coordinator.rs (+100 lines)
- load_learning_profile_from_kg() - core KG integration method
- load_all_learning_profiles() - batch loading for agents
- assign_task() already uses learning-based scoring via AgentScoringService
### Existing Complete Implementation
- vapora-knowledge-graph/src/learning.rs - calculation functions
- vapora-agents/src/learning_profile.rs - data structures and expertise
- vapora-agents/src/scoring.rs - unified scoring service
- vapora-agents/src/profile_adapter.rs - adapter methods
## Tests Passing
- learning_profile: 7 tests ✅
- scoring: 5 tests ✅
- profile_adapter: 6 tests ✅
- coordinator: learning-specific tests ✅
## Data Flow
1. Task arrives → AgentCoordinator::assign_task()
2. Extract task_type from description
3. Query KG for task-type executions (load_learning_profile_from_kg)
4. Calculate expertise with recency bias
5. Score candidates (SwarmCoordinator + learning)
6. Assign to top-scored agent
7. Execution result → KG → Update learning profiles
## Key Design Decisions
✅ Recency bias: 7-day half-life with 3x weight for recent performance
✅ Confidence scoring: min(1.0, total_executions / 20) prevents overfitting
✅ Hierarchical scoring: 30% base load, 50% expertise, 20% confidence
✅ KG query limit: 100 recent executions per task-type for performance
✅ Async loading: load_learning_profile_from_kg supports concurrent loads
## Next: Phase 5.4 - Cost Optimization
Ready to implement budget enforcement and cost-aware provider selection.
2026-01-11 13:03:53 +00:00
|
|
|
|
|
|
|
|
use crate::api::state::AppState;
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Deserialize)]
|
|
|
|
|
pub struct TaskQueryParams {
|
|
|
|
|
pub project_id: String,
|
|
|
|
|
pub status: Option<String>,
|
|
|
|
|
pub assignee: Option<String>,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Deserialize)]
|
|
|
|
|
pub struct ReorderTaskPayload {
|
|
|
|
|
pub task_order: i32,
|
|
|
|
|
pub status: Option<TaskStatus>,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Deserialize)]
|
|
|
|
|
pub struct AssignTaskPayload {
|
|
|
|
|
pub assignee: String,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[derive(Debug, Deserialize)]
|
|
|
|
|
pub struct UpdatePriorityPayload {
|
|
|
|
|
pub priority: TaskPriority,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// List tasks with optional filters
|
|
|
|
|
///
|
|
|
|
|
/// GET /api/v1/tasks?project_id=xxx&status=todo&assignee=agent1
|
|
|
|
|
pub async fn list_tasks(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Query(params): Query<TaskQueryParams>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
let tasks = if let Some(status_str) = params.status {
|
|
|
|
|
// Parse status
|
|
|
|
|
let status: TaskStatus = serde_json::from_value(serde_json::json!(status_str))?;
|
|
|
|
|
state
|
|
|
|
|
.task_service
|
|
|
|
|
.list_tasks_by_status(¶ms.project_id, tenant_id, status)
|
|
|
|
|
.await?
|
|
|
|
|
} else if let Some(assignee) = params.assignee {
|
|
|
|
|
state
|
|
|
|
|
.task_service
|
|
|
|
|
.list_tasks_by_assignee(¶ms.project_id, tenant_id, &assignee)
|
|
|
|
|
.await?
|
|
|
|
|
} else {
|
|
|
|
|
state
|
|
|
|
|
.task_service
|
|
|
|
|
.list_tasks(¶ms.project_id, tenant_id)
|
|
|
|
|
.await?
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
Ok(Json(tasks))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Get a specific task
|
|
|
|
|
///
|
|
|
|
|
/// GET /api/v1/tasks/:id
|
|
|
|
|
pub async fn get_task(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
let task = state.task_service.get_task(&id, tenant_id).await?;
|
|
|
|
|
Ok(Json(task))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Create a new task
|
|
|
|
|
///
|
|
|
|
|
/// POST /api/v1/tasks
|
|
|
|
|
pub async fn create_task(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Json(mut task): Json<Task>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
task.tenant_id = "default".to_string();
|
|
|
|
|
|
|
|
|
|
let created = state.task_service.create_task(task).await?;
|
|
|
|
|
Ok((StatusCode::CREATED, Json(created)))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Update a task
|
|
|
|
|
///
|
|
|
|
|
/// PUT /api/v1/tasks/:id
|
|
|
|
|
pub async fn update_task(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
Json(updates): Json<Task>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
2026-01-11 21:32:56 +00:00
|
|
|
let updated = state
|
|
|
|
|
.task_service
|
|
|
|
|
.update_task(&id, tenant_id, updates)
|
|
|
|
|
.await?;
|
feat: Phase 5.3 - Multi-Agent Learning Infrastructure
Implement intelligent agent learning from Knowledge Graph execution history
with per-task-type expertise tracking, recency bias, and learning curves.
## Phase 5.3 Implementation
### Learning Infrastructure (✅ Complete)
- LearningProfileService with per-task-type expertise metrics
- TaskTypeExpertise model tracking success_rate, confidence, learning curves
- Recency bias weighting: recent 7 days weighted 3x higher (exponential decay)
- Confidence scoring prevents overfitting: min(1.0, executions / 20)
- Learning curves computed from daily execution windows
### Agent Scoring Service (✅ Complete)
- Unified AgentScore combining SwarmCoordinator + learning profiles
- Scoring formula: 0.3*base + 0.5*expertise + 0.2*confidence
- Rank agents by combined score for intelligent assignment
- Support for recency-biased scoring (recent_success_rate)
- Methods: rank_agents, select_best, rank_agents_with_recency
### KG Integration (✅ Complete)
- KGPersistence::get_executions_for_task_type() - query by agent + task type
- KGPersistence::get_agent_executions() - all executions for agent
- Coordinator::load_learning_profile_from_kg() - core KG→Learning integration
- Coordinator::load_all_learning_profiles() - batch load for multiple agents
- Convert PersistedExecution → ExecutionData for learning calculations
### Agent Assignment Integration (✅ Complete)
- AgentCoordinator uses learning profiles for task assignment
- extract_task_type() infers task type from title/description
- assign_task() scores candidates using AgentScoringService
- Fallback to load-based selection if no learning data available
- Learning profiles stored in coordinator.learning_profiles RwLock
### Profile Adapter Enhancements (✅ Complete)
- create_learning_profile() - initialize empty profiles
- add_task_type_expertise() - set task-type expertise
- update_profile_with_learning() - update swarm profiles from learning
## Files Modified
### vapora-knowledge-graph/src/persistence.rs (+30 lines)
- get_executions_for_task_type(agent_id, task_type, limit)
- get_agent_executions(agent_id, limit)
### vapora-agents/src/coordinator.rs (+100 lines)
- load_learning_profile_from_kg() - core KG integration method
- load_all_learning_profiles() - batch loading for agents
- assign_task() already uses learning-based scoring via AgentScoringService
### Existing Complete Implementation
- vapora-knowledge-graph/src/learning.rs - calculation functions
- vapora-agents/src/learning_profile.rs - data structures and expertise
- vapora-agents/src/scoring.rs - unified scoring service
- vapora-agents/src/profile_adapter.rs - adapter methods
## Tests Passing
- learning_profile: 7 tests ✅
- scoring: 5 tests ✅
- profile_adapter: 6 tests ✅
- coordinator: learning-specific tests ✅
## Data Flow
1. Task arrives → AgentCoordinator::assign_task()
2. Extract task_type from description
3. Query KG for task-type executions (load_learning_profile_from_kg)
4. Calculate expertise with recency bias
5. Score candidates (SwarmCoordinator + learning)
6. Assign to top-scored agent
7. Execution result → KG → Update learning profiles
## Key Design Decisions
✅ Recency bias: 7-day half-life with 3x weight for recent performance
✅ Confidence scoring: min(1.0, total_executions / 20) prevents overfitting
✅ Hierarchical scoring: 30% base load, 50% expertise, 20% confidence
✅ KG query limit: 100 recent executions per task-type for performance
✅ Async loading: load_learning_profile_from_kg supports concurrent loads
## Next: Phase 5.4 - Cost Optimization
Ready to implement budget enforcement and cost-aware provider selection.
2026-01-11 13:03:53 +00:00
|
|
|
Ok(Json(updated))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Delete a task
|
|
|
|
|
///
|
|
|
|
|
/// DELETE /api/v1/tasks/:id
|
|
|
|
|
pub async fn delete_task(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
state.task_service.delete_task(&id, tenant_id).await?;
|
|
|
|
|
Ok(StatusCode::NO_CONTENT)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Reorder a task (for Kanban drag & drop)
|
|
|
|
|
///
|
|
|
|
|
/// PUT /api/v1/tasks/:id/reorder
|
|
|
|
|
pub async fn reorder_task(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
Json(payload): Json<ReorderTaskPayload>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
let updated = state
|
|
|
|
|
.task_service
|
|
|
|
|
.reorder_task(&id, tenant_id, payload.task_order, payload.status)
|
|
|
|
|
.await?;
|
|
|
|
|
Ok(Json(updated))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Update task status
|
|
|
|
|
///
|
|
|
|
|
/// PUT /api/v1/tasks/:id/status
|
|
|
|
|
pub async fn update_task_status(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
Json(payload): Json<serde_json::Value>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
let status: TaskStatus = serde_json::from_value(payload["status"].clone())?;
|
|
|
|
|
|
|
|
|
|
let updated = state
|
|
|
|
|
.task_service
|
|
|
|
|
.update_task_status(&id, tenant_id, status)
|
|
|
|
|
.await?;
|
|
|
|
|
Ok(Json(updated))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Assign a task to an agent/user
|
|
|
|
|
///
|
|
|
|
|
/// PUT /api/v1/tasks/:id/assign
|
|
|
|
|
pub async fn assign_task(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
Json(payload): Json<AssignTaskPayload>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
let updated = state
|
|
|
|
|
.task_service
|
|
|
|
|
.assign_task(&id, tenant_id, payload.assignee)
|
|
|
|
|
.await?;
|
|
|
|
|
Ok(Json(updated))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Update task priority
|
|
|
|
|
///
|
|
|
|
|
/// PUT /api/v1/tasks/:id/priority
|
|
|
|
|
pub async fn update_priority(
|
|
|
|
|
State(state): State<AppState>,
|
|
|
|
|
Path(id): Path<String>,
|
|
|
|
|
Json(payload): Json<UpdatePriorityPayload>,
|
|
|
|
|
) -> ApiResult<impl IntoResponse> {
|
|
|
|
|
// TODO: Extract tenant_id from JWT token
|
|
|
|
|
let tenant_id = "default";
|
|
|
|
|
|
|
|
|
|
let updated = state
|
|
|
|
|
.task_service
|
|
|
|
|
.update_priority(&id, tenant_id, payload.priority)
|
|
|
|
|
.await?;
|
|
|
|
|
Ok(Json(updated))
|
|
|
|
|
}
|