Reasoning — Research agent for Claude Code
Implement settled work with sustained analysis of interacting invariants or difficult failure modes.
How to install Reasoning
Installs to ~/.claude/agents/jmcvetta-claude-daily-driver-reasoning.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/jmcvetta/claude-daily-driver/HEAD/agents/reasoning.md -o ~/.claude/agents/jmcvetta-claude-daily-driver-reasoning.md Restart Claude Code, or start a new session, for it to be picked up.
What Reasoning does
name: reasoning description: Implement settled work with sustained analysis of interacting invariants or difficult failure modes. model: "@reasoning"
You are a general-purpose implementation agent. Read the task handoff, inspect relevant repository guidance and patterns, implement the complete requested change, and verify it with focused tests. Use the task worktree for all repository operations. Report concrete changes, checks, and any blocking configuration gap.
Alternatives in Research
- Council Munger — Council member 4.2k ★
- Fable Worker Sonnet — Fable stage worker for tasks needing real reasoning — research synthesis, nontrivial code, analysis, document 837 ★
- Navigator Research — Specialized codebase exploration and architecture discovery 229 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Analysis Correctness
Analysis expert focused on logical correctness, invariants, state consistency, and proof-level reasoning.
Determinism Reviewer
Finds work being done by model reasoning at runtime that tested code could do reliably. Use when auditing a .c
Observability Gap Auditor
Use this agent for adversarial analysis of observability and alerting gaps — silent denial paths, alert deadlo
Communication Excellence Coach
Communication specialist providing email refinement, tone calibration, roleplay practice for difficult convers
Osf Analyze
Codebase structural analysis using codebase-retrieval. Traces dependencies, blast radius, call chains, and imp
Codex Agent
Runs OpenAI Codex on the user's ChatGPT subscription. Three modes - consult (OS-sandboxed read-only analysis,
Related Skills
Verify Before Blocking
A portable discipline that stops AI agents from turning unverified assumptions into stop-work instructions. Si
Bughunt Skill
Agent skill that plays your app or game like real users (seeded personas), checks invariants every step, and t
Deepseek Skill
DeepSeek V4 as a Claude Code skill — five modes including 'audit' that challenges Claude's reasoning in the cu