| name | model-engineer |
| description | Cost-quality optimization agent that analyzes tradeoffs, scores routing candidates, and proposes A/B tests. Use to generate task routing recommendations and experiment designs based on TokenAdvisor findings. |
| license | Proprietary |
| compatibility | Designed for agentic-engineers framework |
| metadata | {"author":"agentic-engineers","version":"1.0","category":"optimization","role":"senior-engineer","model":"sonnet-4-6","effort":"high","schedule":"0 17 * * *"} |
Overview
Model Engineer analyzes TokenAdvisor findings and generates cost-quality optimizations, including task routing recommendations and A/B test proposals.
What it does:
- Analyzes cost-quality tradeoffs by role and model
- Scores task routing candidates (Engineer, Senior, Lead, Principal)
- Generates feedback to TokenAdvisor findings
- Proposes A/B test designs with hypotheses
- Calculates efficiency (cost per quality point)
- Recommends model upgrades and role shifts
Invocation
Manual Run
python scripts/model-engineer.py --analyze
python scripts/model-engineer.py --recommend --complexity medium
python scripts/model-engineer.py --feedback
Automated (Cron)
Daily at 17:15 UTC (15 min after TokenAdvisor) via orchestration/config/model-engineer.cron
0 17 * * * sleep 900 && cd <project-root> && python agentic-engineers/skills/model-engineer/scripts/model-engineer.py --analyze
Configuration
Configuration
- Cost target: $0.0016 per quality point
- Models: Haiku 4.5, Sonnet 4.6, Opus 4.8
- Roles: Engineer, Senior, Lead, Principal
- Input: TokenAdvisor output +
agentic-engineers/data/metrics/
- Output:
agentic-engineers/data/logs/model-engineer-YYYY-MM-DD.log
Integration
Input: TokenAdvisor recommendations + historical metrics
Output: Feeds to A/B Testing Framework (test proposals)
Dashboard: Model Performance (Grafana)
Scripts
model-engineer.py — Main optimization agent (Sonnet 4.6)
Self-Improvement
This skill participates in the framework's continuous improvement cycle
(see skill-improvement-feedback).
When you use model-engineer during a task, include a skill_feedback entry
in your HANDBACK to help improve it over time:
skill_feedback:
- skill_name: model-engineer
effectiveness_score: 0.85
clarity_score: 0.90
coverage_gaps:
- "Specific scenario the skill did not address"
improvement_suggestions:
- "Concrete change that would have helped"
usage_context: "One sentence on how you used this skill"
Positive feedback is as valuable as critical feedback. Three or more
feedback items for this skill automatically trigger an improvement task.
