agent-manager-skill

name	agent-manager-skill
description	Manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling. Includes agent performance optimization workflows.
risk	unknown
source	community

Agent Manager Skill

When to use

Use this skill when you need to:

• run multiple local CLI agents in parallel (separate tmux sessions)
• start/stop agents and tail their logs
• assign tasks to agents and monitor output
• schedule recurring agent work (cron)

Prerequisites

Install agent-manager-skill in your workspace:

git clone https://github.com/fractalmind-ai/agent-manager-skill.git

Common commands

python3 agent-manager/scripts/main.py doctor
python3 agent-manager/scripts/main.py list
python3 agent-manager/scripts/main.py start EMP_0001
python3 agent-manager/scripts/main.py monitor EMP_0001 --follow
python3 agent-manager/scripts/main.py assign EMP_0002 <<'EOF'
Follow teams/fractalmind-ai-maintenance.md Workflow
EOF

Notes

• Requires tmux and python3.
• Agents are configured under an agents/ directory (see the repo for examples).

---

Agent Performance Optimization Workflow

Systematic improvement of existing agents through performance analysis, prompt engineering, and continuous iteration.

Phase 1: Performance Analysis and Baseline

Gather Performance Data

Collect metrics including:

• Task completion rate (successful vs failed tasks)
• Response accuracy and factual correctness
• Tool usage efficiency (correct tools, call frequency)
• Average response time and token consumption
• User satisfaction indicators (corrections, retries)
• Hallucination incidents and error patterns

User Feedback Pattern Analysis

Identify recurring patterns:

• Correction patterns: Where users consistently modify outputs
• Clarification requests: Common areas of ambiguity
• Task abandonment: Points where users give up
• Follow-up questions: Indicators of incomplete responses
• Positive feedback: Successful patterns to preserve

Failure Mode Classification

Categorize failures by root cause:

• Instruction misunderstanding, output format errors
• Context loss in long conversations
• Tool misuse or inefficient tool selection
• Constraint violations, edge case handling

Baseline Performance Report

- Task Success Rate: [X%]
- Average Corrections per Task: [Y]
- Tool Call Efficiency: [Z%]
- User Satisfaction Score: [1-10]
- Average Response Latency: [Xms]
- Token Efficiency Ratio: [X:Y]

Phase 2: Prompt Engineering Improvements

• Chain-of-Thought Enhancement: Add explicit reasoning steps and self-verification checkpoints
• Few-Shot Example Optimization: Curate diverse examples covering common use cases and edge cases
• Role Definition Refinement: Clear mission, expertise domains, behavioral traits, constraints
• Constitutional AI Integration: Self-correction with critique-and-revise loops
• Output Format Tuning: Structured templates, dynamic formatting, progressive disclosure

Phase 3: Testing and Validation

A/B Testing Framework

• Minimum sample size: 100 tasks per variant
• Confidence level: 95% (p < 0.05)
• Compare: success rate, speed, token usage
• Use blind human review + automated scoring

Evaluation Metrics

• Task-Level: Completion rate, correctness, efficiency, tool appropriateness
• Quality: Hallucination rate, consistency, format compliance, safety
• Performance: Latency, token consumption, cost per task

Phase 4: Version Control and Deployment

Version Management

Format: agent-name-v[MAJOR].[MINOR].[PATCH]
MAJOR: Significant capability changes
MINOR: Prompt improvements, new examples
PATCH: Bug fixes, minor adjustments

Staged Rollout

1. Alpha (5% traffic) -> Beta (20%) -> Canary (20-50-100%) -> Full deployment
2. 7-day monitoring period after full deployment

Rollback Triggers

• Success rate drops >10% from baseline
• Critical errors increase >5%
• Cost per task increases >20%
• Safety violations detected

Success Criteria

• Task success rate improves by >=15%
• User corrections decrease by >=25%
• No increase in safety violations
• Response time remains within 10% of baseline

Continuous Improvement Cycle

• Weekly: Monitor metrics and collect feedback
• Monthly: Analyze patterns and plan improvements
• Quarterly: Major version updates with new capabilities