Vapora/scripts/export-tracking.nu
Jesús Pérez d14150da75 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

167 lines
4.4 KiB
Plaintext

#!/usr/bin/env nu
# export-tracking.nu - Export tracking data in multiple formats
# Follows NuShell 0.108+ guidelines with explicit types
def main [
format: string = "json" # Required positional
--output: string = "export" # Flag with value
--project: string = "" # Optional filter
--status: string = "all" # Filter by status
--verbose = false
]: void {
# Rule 3: Early validation
if $format not-in ["json", "csv", "kanban", "markdown"] {
error make {
msg: $"Invalid format: [$format]. Must be: json, csv, kanban, or markdown"
}
}
if $verbose {
print "📊 Starting tracking export..."
print $"📁 Format: [$format]"
if ($project != "") {
print $"🎯 Project filter: [$project]"
}
}
# Rule 13: Predictable naming (get-tracking-data)
let data = get-tracking-data $project $status
# Rule 17: String interpolation
let output-file = $"[$output].($format)"
if $verbose {
print $"📝 Exporting to [$output-file]"
}
# Rule 1: Single purpose - format and save
match $format {
"json" => {
let json-content = ($data | to json)
$json-content | save --force $output-file
}
"csv" => {
let csv-content = format-csv $data
$csv-content | save --force $output-file
}
"kanban" => {
let kanban-content = format-kanban $data
$kanban-content | save --force $output-file
}
"markdown" => {
let md-content = format-markdown $data
$md-content | save --force $output-file
}
}
print $"✅ Exported to [$output-file]"
# Rule 17: ($expr) for expressions
let file-size = (stat $output-file | get size)
print $"📦 File size: (($file-size / 1024) | math round --precision 2) KB"
}
# Rule 1: Single purpose - fetches data
def get-tracking-data [project-filter: string, status-filter: string]: table {
let url = if ($project-filter == "") {
# Rule 17: Expression interpolation
$"http://localhost:3000/api/v1/tracking/summary?status=($status-filter)"
} else {
$"http://localhost:3000/api/v1/tracking/projects/($project-filter)?status=($status-filter)"
}
try {
http get $url | get items
} catch {
print "⚠️ Could not fetch data from tracking API"
print " Make sure tracking server is running: cargo run -p vapora-backend"
[]
}
}
# Rule 1: Single purpose - formats as CSV
def format-csv [data: table]: string {
let header = "id,project,source,type,summary,timestamp\n"
let rows = (
$data
| each { |item|
$"($item.id),($item.project_path),($item.source),($item.entry_type),\"($item.summary)\",($item.timestamp)"
}
| str join "\n"
)
$header + $rows
}
# Rule 1: Single purpose - formats as Kanban
def format-kanban [data: table]: string {
let pending = ($data | where entry_type =~ "Todo" | where status == "Pending")
let in-progress = ($data | where entry_type =~ "Todo" | where status == "InProgress")
let completed = ($data | where entry_type =~ "Todo" | where status == "Completed")
let output = $"
# Kanban Board
## 📋 Pending ($($pending | length) items)
($pending | each { |item|
$"- [$item.summary] *($item.priority)*"
} | str join "\n")
## 🔄 In Progress ($($in-progress | length) items)
($in-progress | each { |item|
$"- [$item.summary] *($item.priority)*"
} | str join "\n")
## ✅ Completed ($($completed | length) items)
($completed | each { |item|
$"- ✅ [$item.summary]"
} | str join "\n")
"
$output
}
# Rule 1: Single purpose - formats as Markdown
def format-markdown [data: table]: string {
let changes = ($data | where source =~ "Change")
let todos = ($data | where source =~ "Todo")
let output = $"
# Tracking Report
Generated: (date now | format date '%Y-%m-%d %H:%M:%S UTC')
## Summary
- **Total Entries**: ($($data | length))
- **Changes**: ($($changes | length))
- **TODOs**: ($($todos | length))
## Changes
($changes | each { |item|
$"### [$item.timestamp] - ($item.summary)
**Impact**: ($item.impact) | **Breaking**: ($item.breaking)
"
} | str join)
## TODOs
($todos | each { |item|
$"### [$item.summary]
**Priority**: ($item.priority) | **Estimate**: ($item.estimate) | **Status**: ($item.status)
"
} | str join)
"
$output
}