zte-progression

Guide progression from In-Loop to Out-Loop to Zero-Touch Engineering. Use when assessing agentic maturity, planning ZTE progression, or identifying requirements for autonomous operation.

⭐3stars🔀0forks📁View on GitHub🕐Updated Jan 3, 2026

When & Why to Use This Skill

The ZTE Progression skill is a strategic framework designed to guide engineering teams through the evolution of agentic workflows. It facilitates the transition from manual 'In-Loop' prompting to 'Zero-Touch Engineering' (ZTE) by providing structured assessments of agentic maturity, identifying technical blockers, and establishing clear KPIs for autonomous operation. This skill is essential for organizations looking to scale AI agent autonomy safely and effectively within their software development lifecycle.

Use Cases

  • Maturity Assessment: Evaluating whether current engineering workflows are In-Loop, Out-Loop, or ready for Zero-Touch automation based on presence KPIs and human intervention frequency.
  • Strategic Roadmap Planning: Developing a step-by-step progression plan to move from manual reviews to fully autonomous 'shipping' for specific task classes like chores or bug fixes.
  • Blocker Identification: Analyzing system weaknesses such as low test coverage, inconsistent success rates, or lack of rollback capabilities that prevent the adoption of ZTE.
  • Confidence Building & KPI Tracking: Monitoring success streaks and review catch rates to provide data-driven justification for increasing agent autonomy.
namezte-progression
descriptionGuide progression from In-Loop to Out-Loop to Zero-Touch Engineering. Use when assessing agentic maturity, planning ZTE progression, or identifying requirements for autonomous operation.
allowed-toolsRead, Grep, Glob

ZTE Progression Skill

Guide teams through the three levels of agentic coding: In-Loop, Out-Loop, and Zero-Touch Engineering.

When to Use

  • Assessing current agentic maturity level
  • Planning progression to higher automation
  • Identifying blockers to ZTE adoption
  • Building confidence for autonomous operation

Core Concepts

Three Levels

Level Presence KPI Description
In-Loop Constant Interactive prompting
Out-Loop 2 (prompt + review) AFK agents with review
Zero-Touch 1 (prompt only) Full automation

Progression Requirements

In-Loop -> Out-Loop:

  • Workflows succeed on first attempt
  • Tests catch issues reliably
  • Similar tasks repeat frequently

Out-Loop -> ZTE:

  • 90%+ success rate
  • Review catches nothing new
  • Tests provide safety net

Assessment Workflow

Step 1: Identify Current Level

Search for agentic workflow indicators:

Look for:
- Manual prompting patterns (In-Loop)
- Automated workflows with review (Out-Loop)
- Fully automated shipping (ZTE)

Step 2: Analyze Success Metrics

Review KPI data if available:

Check @agentic-kpis.md or equivalent:
- Attempt counts
- Success streaks
- Review catch rates

Step 3: Identify Blockers

Common blockers to progression:

Blocker Mitigation
Low test coverage Improve tests before progressing
Inconsistent success Analyze failures, improve prompts
Review catches issues Tests need to catch these first
No rollback capability Add before enabling ZTE

Step 4: Recommend Next Steps

Based on current level, recommend:

At In-Loop:

  • Start with Out-Loop for chores
  • Build workflow automation
  • Establish review process

At Out-Loop:

  • Track review catch rate
  • Build confidence with simple tasks
  • Consider ZTE for high-confidence areas

Near ZTE:

  • Enable for chores first
  • Expand progressively
  • Monitor continuously

Key Memory References

  • @zte-progression.md - Three levels definition
  • @zte-confidence-building.md - Building ZTE confidence
  • @agentic-kpis.md - KPI tracking
  • @composable-primitives.md - Workflow building blocks

Output Format

Provide assessment:

## ZTE Assessment

**Current Level:** [In-Loop | Out-Loop | ZTE]
**Problem Classes Assessed:** [chores, bugs, features]

### Indicators Found
- [Evidence of current level]

### Blockers to Progression
- [Identified blockers]

### Recommendations
1. [Specific next step]
2. [Specific next step]

### Target Timeline
- [Realistic progression timeline]

Anti-Patterns to Identify

  • Skipping levels (Out-Loop before tests exist)
  • All-or-nothing thinking (must ZTE everything)
  • Ignoring failure signals (pushing forward despite issues)
  • Review theater (reviewing without catching anything)

Version History

  • v1.0.0 (2025-12-26): Initial release

Last Updated

Date: 2025-12-26 Model: claude-opus-4-5-20251101