skills.rest
sudhakrms
sudhakrmsCOMMUNITY
Explore all skills by this author →
GitHub

Model Selection

Select language models for agent spawns using a five-layer resolution hierarchy.

Updated Apr 7, 2026
Search Tags:#llm#fallback#model-selection#spawn#cost-aware#log-model
One-click install
npx skills add https://github.com/sudhakrms/baristabro --skill model-selection-sudhakrms
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Model Selection
Source: https://github.com/sudhakrms/baristabro/tree/main/coffee/.squad/templates
Command: npx skills add https://github.com/sudhakrms/baristabro --skill model-selection-sudhakrms

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automatically select the appropriate LLM model for every agent spawn to balance cost, latency, and output quality.

Core Features & Use Cases

  • Layered resolution with a 5-layer hierarchy (Layer 0a, 0b, 1, 2, 3) to determine the model per task.
  • Persistent preferences via .squad/config.json (defaultModel and agentModelOverrides) and session directives that can override those, plus per-charter expectations.
  • Use cases across coding, testing, documentation, triage, and architectural tasks with deterministic model choice and clear logging.

Quick Start

Spawn an agent and observe the coordinator's log showing the resolved model for that spawn.

Frequently Asked Questions about Model Selection

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

FAQPage Schema
How do I automatically select the right LLM for agent spawns to balance cost and quality?▼

Automatic LLM model selection uses a five-layer resolution hierarchy to deterministically pick the best model per agent spawn, balancing cost, latency, and output quality for tasks like coding and triage.

How does deterministic model resolution handle fallbacks when a chosen LLM is unavailable?▼

Deterministic model resolution enforces a fallback chain that activates when a selected LLM is unavailable, ensuring agent spawns continue running while logging the decision and the substituted model.

Can I override the automatic model selection with persistent preferences for specific agents?▼

You can override automatic model selection using persistent preferences defined in `.squad/config.json` via `defaultModel` and `agentModelOverrides`, which the hierarchy honors before applying computed choices.

What is the best way to manage cost-aware model routing across coding, testing, and documentation tasks?▼

Cost-aware model routing is best managed by applying a layered resolution hierarchy that deterministically selects models across coding, testing, docs, triage, and architecture tasks while logging every decision.

Does the model selection mechanism work without external dependencies or components?▼

The model selection mechanism works without external dependencies, using internal layered resolution and session directives to determine models for agent spawns and provide fallbacks when needed.

Why are my agent spawns using a different model than my session directive specified?▼

Agent spawns may use a different model because the five-layer resolution hierarchy evaluates persistent preferences, per-charter expectations, and session directives in a specific order before computing the final model choice.

Related Skills

Explore More Software Engineering Skills→
lbouriezlbouriez

economy-mode

Switch Layer 3 model auto-selection to cost-optimized alternatives when economy mode is active.

Community
Intermediate
quaid-appquaid-app
3

Model Selection

Resolve and persist LLM model choices per agent spawn in .squad/config.json.

Official
Advanced
pacphipacphi
7

swarm-advanced

Coordinate distributed agent workflows across mesh, hierarchical, star, and ring topologies.

Community
Advanced
webmaxruwebmaxru
49

Model Selection

Resolve LLM models per agent spawn using a five-layer hierarchy and persist to .squad/config.json.

Community
Advanced