Autopilot Haiku — Project Management agent for Claude Code
Autopilot worker running on Haiku.
How to install Autopilot Haiku
Installs to ~/.claude/agents/ialias-claudeautopilot-autopilot-haiku.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/iAlias/ClaudeAutopilot/HEAD/agents/autopilot-haiku.md -o ~/.claude/agents/ialias-claudeautopilot-autopilot-haiku.md Restart Claude Code, or start a new session, for it to be picked up.
What Autopilot Haiku does
name: autopilot-haiku description: Autopilot worker running on Haiku. Use only when the autopilot skill delegates a task to this model and effort. model: haiku
You are an autopilot worker. The main session delegated this task to you because your model and effort level fit it best.
You do not see the conversation: work only from the brief you received. It contains the goal, the relevant files and paths, the constraints and the completion criterion.
- Carry the task through to completio
Alternatives in Project Management
- GPT 5.6 Sol — General-purpose agent that runs on GPT-5.6-Sol (routed through the shunt gateway to the ChatGPT/Codex subscrip 74 ★
- Code Review Coordinator — Code review coordinator 63 ★
- Opus Researcher — Use this agent when the user needs deep research, complex analysis, nuanced reasoning, creative problem-solvin 41 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Autopilot Fable High
Autopilot worker running on Fable at high effort. Use only when the autopilot skill delegates a task to this m
Autopilot Fable Low
Autopilot worker running on Fable at low effort. Use only when the autopilot skill delegates a task to this mo
Haiku Executor
Executes a single LOW-effort task from the planner (single file edit, config change, append text, run one comm
Aro:Agent Haiku
Generic task execution agent using Claude 3.5 Haiku model for fast, efficient operations. Ideal for simple tas
Effort Grader
Blind grader for effortmining. Grades one artifact against a task prompt and a fixed rubric ONLY. The input pa
Model Router
Use when dispatching work and unsure which model tier to use — returns a recommended model and effort level ba
Related Skills
Token Save MCP
Stop your coding agent burning context on file reads. Delegates big reads to a cheap worker model and returns
Goal Agent
Dispatch ONE of the DTD-declared agents on a task — agent name validated against the trust contract, model and
Smart Task
Multiuse task command that routes work through a JEV-style triage step into the right effort lane and model ti