What is the primary purpose of Escalation Governance?

It provides a rigorous framework to ensure that moving to a more powerful and expensive model is justified by genuine reasoning requirements rather than trial-and-error or laziness.

How does this skill help reduce AI operational costs?

By identifying 'anti-patterns' like time-pressure escalation or 'maybe it will figure it out' thinking, it keeps tasks running on faster, cheaper models whenever possible.

Can this skill be integrated into automated agent scripts?

Yes, it includes an Agent Schema section allowing agents to declare escalation hints in YAML frontmatter for orchestrators to act upon.

When is it appropriate to escalate from Sonnet to Opus?

Escalation is appropriate for tasks involving high-stakes production deployments, security policy trade-offs, architecture decisions, or resolving deep ambiguities that simpler models cannot parse.

Escalation Governance

Name: Escalation Governance
Author: athola

byathola

•

Productivity & Workflow

Optimizes AI agent workflows by providing a structured framework for deciding when to transition from efficient models to higher-reasoning models.

About

Escalation Governance is a specialized skill designed to manage the trade-off between speed, cost, and reasoning capability in multi-model environments. It enforces a strict 'investigation first' policy, ensuring that transitions from lower-tier models like Claude Haiku or Sonnet to higher-tier models like Opus are only performed when genuinely justified by task complexity. This skill provides agents with a systematic decision framework to identify legitimate triggers—such as security sensitivity or novel pattern implementation—while preventing 'thrashing' and resource waste caused by repetitive failed attempts or lack of root-cause analysis.

Key Features

Categorizes legitimate vs. illegitimate triggers to prevent unnecessary resource consumption.
Includes standardized agent schema hints for security-sensitive and high-stakes tasks.
Enforces 'No Escalation Without Investigation' protocol to eliminate model thrashing.
Provides a systematic 4-step decision framework for evaluating task complexity.
Defines return-to-base protocols to ensure cost efficiency after complex subtasks are completed.
10 GitHub stars

Use Cases

Determining when a security-critical deployment requires the high-fidelity reasoning of a superior model.
Managing token budgets by preventing agents from using Opus for simple debugging tasks.
Orchestrating autonomous agents that need to weigh the cost-benefit of model escalation during long-running tasks.

About

Key Features

Categorizes legitimate vs. illegitimate triggers to prevent unnecessary resource consumption.
Includes standardized agent schema hints for security-sensitive and high-stakes tasks.
Enforces 'No Escalation Without Investigation' protocol to eliminate model thrashing.
Provides a systematic 4-step decision framework for evaluating task complexity.
Defines return-to-base protocols to ensure cost efficiency after complex subtasks are completed.
10 GitHub stars

Use Cases

Determining when a security-critical deployment requires the high-fidelity reasoning of a superior model.
Managing token budgets by preventing agents from using Opus for simple debugging tasks.
Orchestrating autonomous agents that need to weigh the cost-benefit of model escalation during long-running tasks.