ez-stack CLAUDE.md

Project instructions for ez-stack, a tool that helps manage stacked GitHub pull requests, where one change depends on another. They describe the product purpose, design rules, architecture, and guidance for coding agents.

In plain words
What is it for?
Use them when developing ez-stack, making design decisions, or updating the project after important changes or new releases.
Why use it?
They give people and coding agents one shared source of truth for how the project should work. This reduces guesswork and keeps changes aligned with its approach to branches and pull requests.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/rohoswagger/ez-stack/claude-md
Clone the repo
git clone --depth 1 https://github.com/rohoswagger/ez-stack
Per session 5,336 This file is loaded in full into every session.
When invoked 5,336 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.05336 $0.05336
Opus 5 $0.02668 $0.02668
Sonnet 5 $0.01067 $0.01067
Haiku 4.5 $0.00534 $0.00534

Measured 2d ago against content hash 09c57ac2d4f4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ez-stack CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 217 lines

How it starts

The opening of the file, as written. The whole thing — 217 lines — stays where its author put it; the contents beside it link to each section on GitHub.

ez-stack — Project CLAUDE.md

This is a living document. Update it whenever the product mission evolves, a significant design decision is made, a pattern is established, or a new version ships. It is the source of truth for anyone (human or AI) working on this codebase.


Product Mission

ez-stack makes stacked PRs on GitHub effortless, primarily for AI coding agents.

Agents are the primary audience. Humans benefit too, but when there's a design tradeoff, optimize for agents. Every command should be simpler, more intuitive, and more efficient than the raw git/gh equivalent — the goal is to make version control easier for agents, not to expose git's complexity through a different interface.

Every feature should be inherently more useful than the git commands it replaces.


Core Design Principles

  1. Shell out, don't reimplement. ez orchestrates git and gh. It doesn't reimplement git logic. When in doubt, run the real command.

  2. Auto-restack on every mutation. Any command that moves a branch tip (commit, amend, move) automatically restacks children. Users should never have to think about rebasing.

  3. State is append-only metadata. Stack state lives in .git/ez/stack.json — just branch names, parents, parent SHAs, and PR numbers. It is never the source of truth for git state; git is.

  4. Fail loudly with hints. Every error message tells the user what went wrong AND what to do next. No silent failures.

  5. Human output on stderr, machine output on stdout. Colored tree output and spinners go to stderr. JSON output (--json), URLs (ez pr-link), and anything meant to be piped goes to stdout.

  6. Structured exit codes. Agents branch on failure type. Exit codes are documented and stable:

    • 0 = success
    • 1 = unexpected error
    • 2 = GitHub API / gh CLI error
    • 3 = restack blocked — rebase conflict, or one or more branches skipped during a sync/restack (manual resolution required)
    • 4 = stale remote ref
    • 5 = usage error (on trunk, branch not tracked, etc.)
    • 6 = unstaged changes

Read the full file on GitHub · 217 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 217 lines · 5,336 tokens per session scan A 09c57ac2d4f4

Subscribe to this mod's changes

ez-stack CLAUDE.md is an instructions file published in the GitHub repository rohoswagger/ez-stack (23 stars, last pushed 20d ago), licensed MIT. It adds 5,336 tokens to every session, about $0.0267 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.