escalation-l3

Generate an ARCHITECTURE-CONFLICT report and halt the autonomous loop.

Updated Jun 4, 2026
One-click install
npx skills add https://github.com/Kohlex/autopilot-plugin --skill escalation-l3
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: escalation-l3
Source: https://github.com/Kohlex/autopilot-plugin/tree/main/skills/escalation-l3
Command: npx skills add https://github.com/Kohlex/autopilot-plugin --skill escalation-l3

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It surfaces irreconcilable architectural obstacles when L1 and L2 exhaust their lenses, generating ARCHITECTURE-CONFLICT.md and halting the autopilot loop to require human review.

Core Features & Use Cases

  • Generates ARCHITECTURE-CONFLICT.md with a full failure trail, a confident-handoff, and a BLOCKED-ARCH status.
  • Surfaces a HUMAN-DECISION.md to guide operator review and decision-making.
  • Stops the autonomous chain to prevent cascading changes until a human decision is made.

Quick Start

Invoke escalation-l3 after L1 and L2 fail to resolve the obstacle to generate ARCHITECTURE-CONFLICT.md and halt the autopilot loop.

Frequently Asked Questions about escalation-l3

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I handle an architecture conflict when autonomous coding loops hit a dead end?

An architecture conflict requires halting the autonomous loop and generating a formal ARCHITECTURE-CONFLICT.md report. This surfaces a full failure trail and BLOCKED-ARCH status, prompting a confident-handoff to a human operator for review and decision-making.

When do I need to escalate an irreconcilable architectural obstacle to a human operator?

You need to escalate to a human operator when L1 and L2 processing lenses cannot find a viable path within the locked architecture. This prompts a confident-handoff to prevent cascading changes until a human decision is made.

What files are required to generate a formal architecture conflict report?

Generating an architecture conflict report requires ARCHITECTURE.md (locked), capabilities.yaml, SPEC.md, and PLAN.md as inputs. These files provide the context needed to document the failure trail and produce HUMAN-DECISION.md.

How do I stop an autopilot loop from making cascading changes after an architectural failure?

To stop an autopilot loop from making cascading changes, generate an ARCHITECTURE-CONFLICT.md report to set a BLOCKED-ARCH status. This halts the autonomous chain and produces a HUMAN-DECISION.md file to guide operator review.

What is the purpose of a HUMAN-DECISION.md file in software architecture conflict resolution?

A HUMAN-DECISION.md file guides operator review and decision-making during an architecture conflict. It accompanies the ARCHITECTURE-CONFLICT.md report to provide a confident-handoff when the autonomous loop is stopped.

Can I update capabilities.yaml when documenting a blocked architecture status?

Yes, capabilities.yaml is updated as an output when documenting a blocked architecture status. This occurs alongside generating ARCHITECTURE-CONFLICT.md and HUMAN-DECISION.md to reflect the current operational constraints.