197 lines
5.7 KiB
Rust
Raw Normal View History

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
// vapora-agents: NATS message protocol for inter-agent communication
// Phase 2: Message types for agent coordination
use chrono::{DateTime, Utc};
use serde::{Deserialize, Serialize};
/// Agent message envelope for NATS pub/sub
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(tag = "type", rename_all = "snake_case")]
pub enum AgentMessage {
TaskAssigned(TaskAssignment),
TaskStarted(TaskStarted),
TaskProgress(TaskProgress),
TaskCompleted(TaskCompleted),
TaskFailed(TaskFailed),
Heartbeat(Heartbeat),
AgentRegistered(AgentRegistered),
AgentStopped(AgentStopped),
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TaskAssignment {
pub id: String,
pub agent_id: String,
pub required_role: String,
pub title: String,
pub description: String,
pub context: String,
pub priority: u32,
pub deadline: Option<DateTime<Utc>>,
pub assigned_at: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TaskStarted {
pub task_id: String,
pub agent_id: String,
pub started_at: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TaskProgress {
pub task_id: String,
pub agent_id: String,
pub progress_percent: u32,
pub current_step: String,
pub estimated_completion: Option<DateTime<Utc>>,
pub updated_at: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TaskCompleted {
pub task_id: String,
pub agent_id: String,
pub result: String,
pub artifacts: Vec<String>,
pub tokens_used: u64,
pub duration_ms: u64,
pub completed_at: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TaskFailed {
pub task_id: String,
pub agent_id: String,
pub error: String,
pub retry_count: u32,
pub can_retry: bool,
pub failed_at: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct Heartbeat {
pub agent_id: String,
pub status: String,
pub load: f64,
pub active_tasks: u32,
pub total_tasks_completed: u64,
pub uptime_seconds: u64,
pub timestamp: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct AgentRegistered {
pub agent_id: String,
pub role: String,
pub version: String,
pub capabilities: Vec<String>,
pub registered_at: DateTime<Utc>,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct AgentStopped {
pub agent_id: String,
pub role: String,
pub reason: String,
pub stopped_at: DateTime<Utc>,
}
impl AgentMessage {
/// Serialize message to JSON bytes for NATS
pub fn to_bytes(&self) -> Result<Vec<u8>, serde_json::Error> {
serde_json::to_vec(self)
}
/// Deserialize message from JSON bytes
pub fn from_bytes(bytes: &[u8]) -> Result<Self, serde_json::Error> {
serde_json::from_slice(bytes)
}
/// Get message type as string
pub fn message_type(&self) -> &str {
match self {
AgentMessage::TaskAssigned(_) => "task_assigned",
AgentMessage::TaskStarted(_) => "task_started",
AgentMessage::TaskProgress(_) => "task_progress",
AgentMessage::TaskCompleted(_) => "task_completed",
AgentMessage::TaskFailed(_) => "task_failed",
AgentMessage::Heartbeat(_) => "heartbeat",
AgentMessage::AgentRegistered(_) => "agent_registered",
AgentMessage::AgentStopped(_) => "agent_stopped",
}
}
}
/// NATS subjects for agent communication
pub mod subjects {
pub const TASKS_ASSIGNED: &str = "vapora.tasks.assigned";
pub const TASKS_STARTED: &str = "vapora.tasks.started";
pub const TASKS_PROGRESS: &str = "vapora.tasks.progress";
pub const TASKS_COMPLETED: &str = "vapora.tasks.completed";
pub const TASKS_FAILED: &str = "vapora.tasks.failed";
pub const AGENT_HEARTBEAT: &str = "vapora.agent.heartbeat";
pub const AGENT_REGISTERED: &str = "vapora.agent.registered";
pub const AGENT_STOPPED: &str = "vapora.agent.stopped";
/// Get subject for a specific agent role
pub fn agent_role_subject(role: &str) -> String {
format!("vapora.agent.role.{}", role)
}
/// Get subject for a specific task
pub fn task_subject(task_id: &str) -> String {
format!("vapora.task.{}", task_id)
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_message_serialization() {
let msg = AgentMessage::TaskAssigned(TaskAssignment {
id: "task-123".to_string(),
agent_id: "agent-001".to_string(),
required_role: "developer".to_string(),
title: "Test task".to_string(),
description: "Test description".to_string(),
context: "{}".to_string(),
priority: 80,
deadline: None,
assigned_at: Utc::now(),
});
let bytes = msg.to_bytes().unwrap();
let deserialized = AgentMessage::from_bytes(&bytes).unwrap();
assert_eq!(msg.message_type(), deserialized.message_type());
}
#[test]
fn test_heartbeat_message() {
let heartbeat = Heartbeat {
agent_id: "agent-001".to_string(),
status: "active".to_string(),
load: 0.5,
active_tasks: 2,
total_tasks_completed: 100,
uptime_seconds: 3600,
timestamp: Utc::now(),
};
let msg = AgentMessage::Heartbeat(heartbeat);
assert_eq!(msg.message_type(), "heartbeat");
}
#[test]
fn test_subject_generation() {
assert_eq!(
subjects::agent_role_subject("developer"),
"vapora.agent.role.developer"
);
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
assert_eq!(subjects::task_subject("task-123"), "vapora.task.task-123");
}
}