Loop Deep Reasoner — Development agent for Claude Code
Difficult architecture or unresolved reasoning needing Opus.
How to install Loop Deep Reasoner
Installs to ~/.claude/agents/rokas-capture-build-loop-loop-deep-reasoner.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/rokas-capture/build-loop/HEAD/agents/loop-deep-reasoner.md -o ~/.claude/agents/rokas-capture-build-loop-loop-deep-reasoner.md Restart Claude Code, or start a new session, for it to be picked up.
What Loop Deep Reasoner does
name: loop-deep-reasoner description: Difficult architecture or unresolved reasoning needing Opus. model: opus effort: high disallowedTools: Agent
Resolve only the hard reasoning problem in the parent's brief using the evidence already gathered. Do not repeat broad research, manually bootstrap the EA brain, or delegate. State assumptions, meaningful options, recommended decision, and remaining uncertainty. Do not change state unless explicitly assigned. Keep high effort; xhigh/max/Fable
Alternatives in Development
- Django Pro — You are a Django expert specializing in Django 5.x best practices, scalable architecture, and modern 31.9k ★
- Review Bugs — Reviews code changes for potential bugs found by single-threaded sequential reasoning: logic errors, null safe 432 ★
- Finding Checker — Blind adversarial checker — given ONLY a finding artifact and its evidence (never the author's reasoning), tri 348 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Deep Worker
Opus-tier worker for subtasks that need hard reasoning: root-cause a stubborn bug, evaluate an architecture op
Executor Sonnet Xhigh
Ticket executor — Sonnet at x-high reasoning effort. Rarely the right pairing (a ticket needing xhigh usually
Jev Deep
Repository worker pinned to the deep tier (opus). Use when you already know the tier you want and would rather
Fleet Implement Deep
Hard implementation work: multi-file changes, tricky algorithms, or migrations that need deep reasoning. Runs
Ontology Reasoner
Use this agent when you need to validate the logical consistency of an ontology, check for reasoning errors, i
Premium Reasoner
Top-tier scoped reasoning on your highest-capability model. Use ONLY when the human explicitly authorized prem
Related Skills
Vaos Voice Bridge
Voice bridge connecting PersonaPlex (System 1 fast response) with Letta/Claude (System 2 reasoning). Talker-Re
Multi Reasoner
ChatGPT reasoning MCP for Claude Code - uses Codex CLI, no API billing
Twin Sparrow Deep Audit
Composite deep audit skill combining expert-engineer, code-quality-skill, and oppus-reasoning-contract for str