npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

pitway

v1.1.1

Published

A controlled workflow for agentic software development.

Readme

PitWay

The pit crew for agentic coding.
A controlled workflow for agentic software development.

npm version npm downloads codecov License: MIT

PitWay is an npm-distributed CLI that controls the engineering process around AI coding agents — it is not itself an agent.

  • Agents drive the interaction
  • PitWay controls workflow state, engineering boundaries, verification, and traceability

Why PitWay?

AI coding agents move fast — and that's exactly the problem. Left unstructured, an agent can drift from what was asked, skip verification, or quietly touch files outside its intended scope, with no durable record of what actually happened or why.

PitWay doesn't replace the agent driving your work — it gives that work a process:

  • A confirmed plan before code. Every milestone starts as a contract — objective, acceptance criteria, verification checks — reviewed and approved by a human before any implementation begins.
  • Boundaries an agent can't quietly cross. write_scope mechanically limits what a task may touch.
  • Verification, not vibes. Every task is checked against its own declared command before it's considered done, with a mandatory full test suite gating milestone completion.
  • A record that survives the conversation. Git commits carry traceable PitWay-Milestone/PitWay-Task trailers — the history holds even after the AI session that produced it is gone.

The result: agents move fast, and the engineering process stays in control of where they land.


How It Works

PitWay workflow: BRS/Backlog into Milestone (Contract ⇄ Milestone Review) through the Human Approval gate to the Task Graph, TDD → Task Verification → Task Commit repeating with a Backlog exit, then Final Full Test (failure loops through milestone revision), Milestone Complete, and an opt-in Quick Change lane for small bounded fixes when no milestone is active (TDD → Verify → Human Approval → Commit), ending at Milestone Merge

Source: docs/assets/workflow.mmd (Mermaid) — rendered to SVG so it displays on npmjs.com too, which doesn't render Mermaid.

Workflow Lifecycle:

RequirementMilestone (ContractMilestone Review) → Human ApprovalTask Graph → [TDDTask VerificationTask Commit]* → Final Full Test (fail ⇒ revision loop) → Milestone CompleteMilestone Merge — plus a Quick Change lane for small bounded fixes whenever no milestone is active.

  • Workflow Enforcement: Validates state transitions and enforces task write boundaries.
  • Two-Tier Verification: Runs targeted verification for each task, followed by a mandatory full test suite before closing the milestone.
  • Durable State: Creates traceable Git checkpoints without relying on transient AI conversation memory.

Which Workflow Should I Use?

  • A new capability, not started yet → a Milestone (contract, task graph, human approval).
  • Work discovered while a milestone is active → a Task if it belongs in the graph, or the Backlog if it doesn't.
  • A small, independent fix, no milestone active → a Quick Change — bounded, single-commit.
  • Running work in parallel → an execution mode (execution.strategy: parallel_worktrees), not a separate lane — it applies within whichever of the above you're already using.

See USAGE.md for the full walkthrough of each.


Quickstart

1. Install & Initialize

Run once from the root of a Git repository (run git init first if needed):

npm install -g pitway

pitway init
  • PitWay initializes .pitway/ and installs the Claude Code integration by default.
  • Use pitway init --no-claude to opt out of the Claude Code integration.
  • Use pitway init --opencode to also install the OpenCode integration (.opencode/ — commands, skills, and driver protocol documents).
  • Use pitway init --codex to also install the Codex integration (.codex/ — commands, skills, and driver protocol documents).
  • Upgraded pitway? Run pitway init --reconfigure. A plain re-run of pitway init only fills in files that don't exist yet — it never updates ones you already have, so installed commands/skills/protocol docs silently go stale after an upgrade. --reconfigure refreshes every managed integration asset to the newly installed version; .pitway/ state (milestones, contracts, tasks) is always preserved.
  • init also creates root AGENTS.md/CLAUDE.md instruction files. PitWay's content lives inside an explicit <!-- pitway:managed:start/end --> block — if you already have your own AGENTS.md or CLAUDE.md, the block is appended and your content is left intact; PitWay only ever owns the marked block.

2. Resume Workflow

Inspect or continue the current workflow at any time:

pitway resume

📌 Commit Traceability: Before implementation begins, the developer reviews and confirms the milestone contract. Task commits carry PitWay-Milestone and PitWay-Task Git trailers; milestone baseline and completion commits carry the milestone trailer.

📊 Progress at a Glance: Once a milestone is confirmed, routine driver updates end with a one-line progress footer (e.g. 🏎️ 54% · ✅ 7/12 · Next: T008). pitway milestone-status [id] renders the full status report — workload, task table, critical path, token breakdown, and a racing footer with progress bar — for the active milestone by default, or an explicit id for any milestone.

3. Explore Commands

Run the following for the full, authoritative CLI command surface and available flags:

pitway --help

For a hands-on walkthrough of the whole workflow — drafting a contract, confirming a milestone, working a task through to completion — see USAGE.md.

💡 Driver Support: Claude Code (installed by default), OpenCode (opt-in), and Codex (opt-in) driver integrations ship as text assets from a shared common layer — skills and protocol documents are defined once and resolved per driver. PitWay's Core remains provider-agnostic.

⚠️ What PitWay enforces vs what relies on driver discipline. Mechanically enforced: the state machines, write_scope boundaries, verification gates, commit trailers, and git-safety checks — no driver can bypass these through the CLI. Installed-instruction-only: stopping for human approval gates, driver-presented progress footers, and bounded worker reports are mandated by the installed protocol documents every driver loads, but PitWay cannot observe a live session's obedience — violations surface in review/audit, never at runtime.


Commands & Integration

  • Command Reference: Run pitway --help for the full, authoritative CLI command surface and available flags. The milestone-* commands also answer to shorter ms-* aliases (pitway ms-status, ms-confirm, ms-merge, …).
  • Usage Guide: See USAGE.md for a hands-on walkthrough — installation, your first milestone end to end, inspecting state, mid-flight corrections, and a full command reference table.
  • Claude Code: pitway init installs PitWay's commands as real Claude Code slash commands (.claude/commands/*.md, each carrying description/argument-hint metadata for the / picker), alongside the driver protocol documents that explain how and when to use them.
  • OpenCode: pitway init --opencode installs the same command surface in OpenCode's own convention (.opencode/commands/*.md) plus the shared skills and protocol documents. Skills and protocol content come from the common layer — defined once, never forked per driver.
  • Codex: pitway init --codex installs the same command surface in Codex's convention (.codex/commands/*.md) plus the shared skills and protocol documents. Skills and protocol content come from the common layer — defined once, never forked per driver.

Engineering Boundaries

| Boundary Property | Behavior & Scope | | --- | --- | | write_scope | Mechanically enforced task boundary to prevent unintended file modifications. | | context_files | Limits task-context bundles supplied by PitWay (not an OS-level read sandbox). | | Agent Runtime | PitWay does not claim control over external agent runtimes, shells, or OS tool permissions. | | Milestone Review | Reviewers produce findings only. PitWay does not run reviews or verify reviewer independence. |


Workflow Policies

Two repository-level policies live in .pitway/config.yaml. pitway init generates them with the recommended workflow enabled (with explanatory comments in the file):

git:
  branch_strategy: milestone     # each milestone gets its own dedicated branch
execution:
  strategy: parallel_worktrees   # independent tasks dispatch concurrently
  • git.branch_strategy: main | milestonemilestone (the generated default) gives each milestone its own dedicated branch, checked out for the milestone's full lifecycle; once completed, pitway milestone-merge <id> lands the branch into its base branch with full git-safety checks and idempotent re-runs. Set main to commit milestones directly to the current branch instead.
  • execution.strategy: sequential | parallel_worktreesparallel_worktrees (the generated default) lets independent, dependency-free, disjoint-write_scope tasks dispatch concurrently, each into its own temporary Git worktree; PitWay validates eligibility and integrates each result as a diff-apply — never a merge — so the resulting mainline history stays indistinguishable from sequential execution. Set sequential to run one task at a time, inline.

Dogfooding & Verification

PitWay is developed and maintained using its own workflow:

  • M001–M003: Bootstrap foundation
  • M004: Crossed the self-hosting boundary
  • M005+: Created, verified, and completed entirely through PitWay

All project claims are bounded strictly by evidence that survives a fresh clone: committed Git history, .pitway/ state, and the automated test suite.


Maintenance & Security

Dependency updates arrive as Dependabot pull requests (weekly, grouped dev-dependencies, no auto-merge — see .github/dependabot.yml). Dependabot security alerts are a repository setting, not something this file turns on: enable them under Settings → Code security if you want them.

Static analysis runs via CodeQL (.github/workflows/codeql.yml), scanning the TypeScript/JavaScript source for common vulnerability patterns. Follows GitHub's default cadence: every push and pull request to main, plus a weekly schedule. Results appear under the repository's Security → Code scanning tab.

See SECURITY.md for how to report a vulnerability.


Release History

For changes and improvements introduced in each published version, see CHANGELOG.md.


There is a way to build with agents. This is PitWay.

License

MIT — see LICENSE.