Pilot Shell enhances Claude Code with structured workflows for spec-driven development, efficient bug fixes, and product requirement generation. It integrates quality gates, preserves context, and optimizes token usage, enabling engineers to produce consistent, production-grade code.

Is pilot-shell open source?

Yes, licensed under NOASSERTION.

pilot-shell

Active·★ 2.1k·NOASSERTION·Updated 2026-10-27

★ Trending

Pilot Shell is installed on top of Claude Code and adds a structured workflow through commands, rules, hooks, skills, subagents, MCP and settings. The headline command is `/spec`, which replaces Claude Code's built-in plan mode for new features and refactors with a flow of discuss, plan, approve, implement with TDD, verify and loop until done; the README also describes teammates annotating a shared spec link. `/fix` is a bugfix workflow with TDD that bails out when the problem is too complex for the standard lane, `/prd` turns ideas into requirements with optional research, and `/setup-rules` generates project rules. The README's pitch is that Claude Code is fast but skips tests and loses context without structure.

Install with `curl -fsSL https://raw.githubusercontent.com/maxritter/pilot-shell/main/install.sh | bash` on macOS, Linux or Windows via WSL2; tools and rules go to `~/.pilot/` and `~/.claude/`, then you run `pilot` or `ccp` inside a project. The README says to use Claude Code's native installer first, recommends a Claude Max, Team or Enterprise plan for heavy use, and documents downgrading to a previous version, uninstalling and resetting. The README is long (over 50k characters). GitHub's license detection shows none, so check the repository for the actual terms.

#ai-agents#ai-assistant#ai-coding#ai-coding-tools#ai-engineering#ai-tools#anthropic#anthropic-claude

Features

01Spec-Driven Development with /spec for end-to-end feature planning and verification.

02Streamlined TDD-driven bugfix workflow with /fix for efficient bug resolution.

03Enforced quality gates with linting, formatting, type checking, and tests.

04Persistent knowledge and context engineering across sessions.

05Token optimization for 60-90% cost reduction.

Why choose it

+Gives Claude Code a clear plan, implement (TDD), verify loop instead of ad-hoc prompting

+Separate workflows for features (`/spec`), bugs (`/fix`) and requirements (`/prd`)

+Spec review can be shared with teammates through a link

+Documented install, downgrade, uninstall and reset steps

Trade-offs

!Only works on top of Claude Code, and the README recommends a paid Claude plan for heavy use

!Installation pipes a remote script into bash and writes to `~/.claude/`; read the script first

!Heavier process than plain Claude Code; small tasks may not need a full spec flow

!License is not detected by GitHub; confirm the terms before using it at work

Maintenance check

ActiveChecked: 2026-10-25

Last commit: 2026-10-25
Commits, last 90 days: 100+
Releases, last 90 days: 30
Latest release: v12.2.0 (2026-10-24)
Open issues: 4
Recent issues with at least one comment: 28/30

Commits per week, last 12 weeks

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

#206 Unclear what to do2 comments · 2026-10-25
#209 Feedback in the app if it is still working or doing something1 comments · 2026-10-25
#207 Deleting a session should stop it's process1 comments · 2026-10-25
#210 Big file diffs do not load in the UI0 comments · 2026-10-25

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

Who it fits

Pick it if

You use Claude Code and want `/spec` to replace plan mode with discuss, plan, approve, TDD, verify.
You want separate flows for bugs (`/fix`) and requirements (`/prd`).
You want teammates to review a spec through a shared link.

Look elsewhere if

You use an agent other than Claude Code: it only works on top of Claude Code.
You need clear licence terms: GitHub does not detect a licence, so confirm before using it at work.
Most of your tasks are small: a full spec flow is heavier than plain Claude Code.

How it differs

vs ouroboros★ 6.3k

Ouroboros is spec-first across many runtimes, with an interview and evaluation gate. Pilot Shell is tied to Claude Code and built around TDD.

Full comparison →

Compatibility

macOS

Verified via docs

Linux

Verified via docs

Windows

Windows (WSL2)

Verified via docs

Quick start

$ curl -fsSL https://raw.githubusercontent.com/maxritter/pilot-shell/main/install.sh | bash

Use cases

↳Structured development of new features, refactoring, and architectural work with TDD.

↳Efficiently investigate and fix bugs with a TDD-driven workflow.

↳Brainstorming and refining vague ideas into clear product requirement documents.

Alternatives

⚡Top VS:

VSouroboros★6.3k→VSclaude-code-ultimate-guide★6.3k→VSentroly★480→VSagents-best-practices★2.4k→

ouroboros★ 6.3k

Agent OS: Stop prompting. Start specifying. A Socratic interview gates the spec on an ambiguity score, then one command drives execution, a 3-stage evaluation gate, and a budgeted evolution loop. MCP server, 13 runtimes: Claude Code, Codex CLI, Gemini CLI, OpenCode, Copilot, Kiro and more.

Features

Why choose it

Trade-offs

Maintenance check

Most discussed open issues

Who it fits

How it differs

Compatibility

Quick start

Use cases

Alternatives

Featured in these lists

Comments

Features

Why choose it

Trade-offs

Maintenance check

Most discussed open issues

Who it fits

How it differs

Compatibility

Quick start

Use cases

Alternatives

Featured in these lists

Comments