How do I install the Fleet Doctrine skill?

Run npx skillmds@latest add dvcrn/fleet-doctrine in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Fleet Doctrine safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Fleet Doctrine?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Fleet Doctrine free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Fleet Doctrine?

dvcrn (@dvcrn) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Fleet Doctrine by dvcrn — Agent Skill

Name: Fleet Doctrine
Rating: 4.5 (4 reviews)
Author: dvcrn

Fleet Doctrine

Model routing strategy for multi-model AI fleet. Use when spawning sub-agents, choosing models for cron jobs, delegating coding tasks, or deciding which model should handle a task. Covers Opus, Codex, Sonnet, Gemini, and Grok routing rules.

dvcrn Updated Sep 8, 2026 32 repo stars

File contents

Fleet Doctrine — Model Routing

Aliases

opus → anthropic/claude-opus-4-6
sonnet → anthropic/claude-sonnet-4-6
codex → openai-codex/gpt-5.3-codex
Gemini: google/gemini-3-pro-preview
Grok: xai/grok-4

Routing Rules

Opus — Commander

When: Main session, orchestration, security decisions, financial tasks, reviewing other models' output, anything high-stakes or ambiguous. Never waste on: Routine crons, simple lookups, email summaries, templated tasks.

Codex — Chief Engineer

When: Big coding tasks (refactors, new features, full repo work), PR reviews, debugging complex issues, checking other models' code output. Spawn as: sessions_spawn(model: "codex", task: "...") or sub-agent with --model codex. Pairs with: Grok for parallel work or second opinions on code.

Sonnet — Workhorse

When: Cron jobs, email briefings, admissions reports, routine admin, quick lookups, drafts, form letters, anything repetitive or templated. Default for: All crons unless the task requires reasoning.

Gemini — Creative & Vision

When: Image generation, analysing long documents (1M context), visual tasks, when a different perspective helps. Best at: Processing massive context windows, multimodal input.

Grok — Fast Backup

When: Parallel work alongside Codex, speed-over-depth tasks, sanity-checking other models' output, when you need a quick second opinion. Good for: Lightweight code reviews, fast research, draft generation.

Decision Flow

Is it security, finance, or high-stakes? → Opus
Is it a big coding task or repo work? → Codex (with Grok as backup)
Does it need image gen or 1M+ context? → Gemini
Is it routine/scheduled/templated? → Sonnet
Need a fast second opinion? → Grok
Not sure? → Opus decides, then delegates.

Fallback Rule

If a model is unavailable on your instance, fall back to your default model (typically Sonnet). The doctrine describes intent — use the best model you have access to that matches the category. When in doubt, do the task with what you've got rather than failing.

Anti-Patterns

Don't use Opus for email summaries or cron jobs
Don't use Sonnet for complex multi-step reasoning
Don't use Codex for non-coding tasks
Don't spawn multiple models on the same task unless deliberately seeking a second opinion

---name: fleet-doctrinedescription: Model routing strategy for multi-model AI fleet. Use when spawning sub-agents, choosing models for cron jobs, delegating coding tasks, or deciding which model should handle a task. Covers Opus, Codex, Sonnet, Gemini, and Grok routing rules.---# Fleet Doctrine — Model Routing## Aliases- `opus` → `anthropic/claude-opus-4-6`- `sonnet` → `anthropic/claude-sonnet-4-6`- `codex` → `openai-codex/gpt-5.3-codex`- Gemini: `google/gemini-3-pro-preview`- Grok: `xai/grok-4`## Routing Rules### Opus — Commander**When:** Main session, orchestration, security decisions, financial tasks, reviewing other models' output, anything high-stakes or ambiguous.**Never waste on:** Routine crons, simple lookups, email summaries, templated tasks.### Codex — Chief Engineer**When:** Big coding tasks (refactors, new features, full repo work), PR reviews, debugging complex issues, checking other models' code output.**Spawn as:** `sessions_spawn(model: "codex", task: "...")` or sub-agent with `--model codex`.**Pairs with:** Grok for parallel work or second opinions on code.### Sonnet — Workhorse**When:** Cron jobs, email briefings, admissions reports, routine admin, quick lookups, drafts, form letters, anything repetitive or templated.**Default for:** All crons unless the task requires reasoning.### Gemini — Creative & Vision**When:** Image generation, analysing long documents (1M context), visual tasks, when a different perspective helps.**Best at:** Processing massive context windows, multimodal input.### Grok — Fast Backup**When:** Parallel work alongside Codex, speed-over-depth tasks, sanity-checking other models' output, when you need a quick second opinion.**Good for:** Lightweight code reviews, fast research, draft generation.## Decision Flow1. Is it security, finance, or high-stakes? → **Opus**2. Is it a big coding task or repo work? → **Codex** (with Grok as backup)3. Does it need image gen or 1M+ context? → **Gemini**4. Is it routine/scheduled/templated? → **Sonnet**5. Need a fast second opinion? → **Grok**6. Not sure? → **Opus** decides, then delegates.## Fallback RuleIf a model is unavailable on your instance, fall back to your default model (typically Sonnet). The doctrine describes *intent* — use the best model you have access to that matches the category. When in doubt, do the task with what you've got rather than failing.## Anti-Patterns- Don't use Opus for email summaries or cron jobs- Don't use Sonnet for complex multi-step reasoning- Don't use Codex for non-coding tasks- Don't spawn multiple models on the same task unless deliberately seeking a second opinion

Frequently asked questions