Skip to content

aalmada/Stationary / .agents/skills/model-selection

model-selection

Verified

Choose and tune which AI model a custom agent, subagent, or squad node should use. USE FOR: picking a model by task capability tier (general-purpose coding/writing, fast/cheap, deep reasoning/debugging, multimodal); efficiency-first vs. capability-first starting strategies; tuning an effort/reasoning-effort parameter instead of switching models; call cadence vs. provider prompt-cache TTL; cost-tier ranking and cost ceilings; cross-provider or cross-model-family diversity for independent/adversarial review; per-harness model field syntax and resolution order (GitHub Copilot CLI, Claude Code, VS Code custom agents); extending this framework to a new harness. DO NOT USE FOR: authoring the rest of an agent's frontmatter or prompt body (use copilot-cli-custom-agents, claude-code-custom-agents, or the target harness's own skill); designing squad topology (use agent-graph-orchestration); SKILL.md authoring (use create-skill).

Model Selection

Every custom agent, subagent, or squad node needs an explicit model choice — never leave it unset on the assumption that an inherited default is adequate for a specialist role. The right model balances five criteria, then gets mapped onto whatever field syntax the target harness actually accepts.

Anatomy

CriterionQuestion it answers
Capability tierHow complex is the reasoning this agent's task requires?
SpeedHow latency-sensitive is whoever is waiting on this agent?
CostWhat's the budget, and is this agent bounded by a caller's own cost ceiling?
Context & modalityDoes the task need a large context window, or image/visual input?
EffortCan the same model be tuned cheaper or faster instead of switching models entirely?

Workflow

  1. Establish task criteria for this specific agent — capability need, latency sensitivity, cost budget, context/multimodal needs → references/selection-criteria.md
  2. Pick a starting strategy: efficiency-first (start cheap/fast, upgrade only if evals show a gap) or capability-first (start at the top, optimize down later) → references/selection-criteria.md
  3. Prefer tuning an effort/reasoning-effort parameter over switching models when the harness and model support it — it trades intelligence for latency/cost within the same model family, a cheaper lever than a full model swap → references/selection-criteria.md
  4. If this agent is one node of a multi-agent squad, or needs independent/adversarial review, apply call-cadence vs. provider prompt-cache TTL, cost-tier ranking/ceilings, and cross-provider diversity → references/cache-and-cost.md
  5. Ask which harness targets this agent (if not already established), then apply that harness's exact model field name, value shape, and resolution order → references/harness-mapping.md
  6. Verify the chosen model name is still current against the provider's live docs before finalizing — model lineups rotate every few months; never trust a memorized or previously-seen model name as still accurate → references/selection-criteria.md
  7. Run markdown-best-practices over any new or edited Markdown produced while documenting the choice (design notes, agent files' surrounding docs)

Reference Files

FileLoad When
references/selection-criteria.mdPicking a capability tier for a task, choosing an efficiency-first vs. capability-first starting point, tuning an effort/reasoning-effort parameter, deciding whether to upgrade/downgrade after benchmarking, or verifying a model name is still current
references/cache-and-cost.mdAn agent's call cadence vs. its provider's prompt-cache TTL, ranking candidate models by cost tier, applying a cost ceiling inherited from a caller, or picking genuinely independent models/providers for adversarial or second-opinion review
references/harness-mapping.mdWriting the actual model:-equivalent field for GitHub Copilot CLI, Claude Code, or VS Code custom agents, understanding each harness's resolution order and fallback-list support, or extending this skill's guidance to a harness not yet covered
brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 319kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 196kUpdated Claude CodeCodexVerifiedInstall