Skip to content

LeapNUX/5nux

Repository files navigation

5-NUX

A free, open-source project-management toolkit you run from the terminal — built for software teams that have to prove their work to auditors, regulators, or compliance officers. It auto-generates the paperwork those reviewers ask for, straight out of your git repo.

5-NUX gives you a whole tree. You provide the soil and you ship yourself.

License: Apache-2.0 Node: >=20 Tests: 587 passing Version: v0.5.0-alpha.1 AI-agent native


A quick glossary (read this first)

5-NUX touches a few areas with their own jargon. Here's the plain-English version we'll use throughout.

The tool names — don't be put off by them. The five tools are named after parts of a tree (root, trunk, branch, leaf, fruit) because each one maps to a stage of a project's life. You don't have to memorize them — just remember what each tool does:

Tool name Just think of it as… What it does in one line
rootnux the Planning tool Writes specs, decision records, risk register, knowledge base
trunknux the Sprint tool Sets up sprint folders and weekly logs
branchnux the Test tool Drafts test plans and runs reports
leafnux the Health-Check tool Red/amber/green snapshot of your project
fruitnux the Audit-Pack tool Generates the paperwork auditors ask for
5nux the Bundle One install that gives you all five tools
6nux-core the Shared library Internal — you don't run it directly

The audit and compliance jargon:

Term you'll see What it actually means
Regulated software Anything that has to pass audits — finance, health-tech, government, anything bound by SOC 2, FedRAMP, HIPAA, ISO, etc.
RTM (Requirements Traceability Matrix) A spreadsheet that links each requirement to the code that implements it and the test that proves it. Auditors ask for this constantly.
SCA (Security Control Assessment) A short report that says "here's a security control, here's how we implemented it, here's the evidence it works."
OSCAL A machine-readable format (JSON) the U.S. government uses for compliance reports. Auditors at FedRAMP and SOC 2 firms can ingest it directly.
HMAC sign-off A cryptographic signature on each entry, chained together so any later edit is detectable. Think "tamper-proof receipt."
ADR (Architecture Decision Record) A one-page note explaining why you made a key technical choice (e.g. "we picked Postgres over Mongo because…").
[VERIFY] marker A flag on AI-drafted text that means "a human still has to read this and confirm it before it counts as verified."
Verb A subcommand you run, like git commit or npm install. Each 5-NUX tool is a small CLI with a handful of verbs.

What 5-NUX does for your project

It runs the full project paper trail — requirement → sprint → test → check → audit hand-off — entirely from the terminal, as plain files in your git repo. No SaaS subscription, no login, no vendor you can't replace.

Two things make it different from every other open-source project tool:

  1. Built for teams that have to pass audits. It auto-generates the documents reviewers actually ask for — traceability matrices, security control reports, machine-readable compliance JSON, tamper-proof sign-off ledgers. It's the open-source alternative to enterprise tools like IBM DOORS, Polarion, Jama, and codeBeamer (which charge $1k–$5k+ per seat per year and lock your data inside their database). See docs/scope.md for the full comparison.
  2. Designed to work alongside AI coding assistants. Every artifact is a plain text file, every command produces machine-readable output (--json), so an AI agent can drive 5-NUX the same way a human can — no special API, no auth dance. A direct Claude Code integration (Model Context Protocol — Anthropic's standard for letting AI assistants call external tools) is coming in v0.6+; today you can already invoke any verb via npx branchnux <command>. See docs/collaboration.md for how AI and humans split the work.

What you actually get:

What you want The command What lands in your repo
Regenerate the traceability matrix (every requirement ↔ its code ↔ its tests) fruitnux rtm (audit-pack tool) requirements/TRACEABILITY.md
An auditor-ready security control report (8 standard sections) fruitnux sca generate <surface> sca/<surface>.md
The same report in the U.S. government's machine-readable JSON format fruitnux sca oscal <surface> OSCAL 1.1.2 JSON file
A cryptographically signed PDF — any later tampering is detectable fruitnux sign <surface> Signed PDF + sign-off ledger
A "why we chose this" decision record, auto-numbered rootnux adr-new <title> (planning tool) docs/adr/NNNN-<slug>.md
A new entry in your risk register rootnux risk-add row appended to requirements/risks/risks.md
A dated sprint folder with starter README and log trunknux new-sprint <slug> (sprint tool) sprint-log/<date>_<slug>/
An AI-drafted test plan (every AI line flagged for human review) branchnux plan <slug> (test tool) testing-log/<date>_<slug>/test-plan.md
A red/amber/green snapshot of your whole project leafnux health (health-check tool) terminal output (or JSON)

Is 5-NUX by itself enough to ship and pass audits? Yes. Other tools (kanban boards, team chat, dashboards, CI/CD) are nice pairings, not requirements.

What 5-NUX doesn't include — and your three options

For each capability outside 5-NUX, you have three choices:

If you want… A: Use an existing app B: Build it yourself C: Pay for the LeapNuX hosted version
Task tracking + kanban GitHub Issues, Linear, Jira, Asana, Trello A custom kanban over the markdown files LeapNuX hosted board
Roadmap / Gantt timeline Productboard, Aha!, GitHub Projects Render a Gantt from sprint folder dates LeapNuX roadmap view
Team chat + notifications Slack, Discord, Microsoft Teams Self-host Mattermost / Rocket.Chat LeapNuX notification hub
Build + deploy pipelines GitHub Actions, CircleCI, Jenkins, Vercel Self-host Drone / Concourse Out of scope (use existing tools)
A nice GUI for non-technical stakeholders (compliance, executives, board) None that show traceability and security reports natively Render the --json output into your own dashboard LeapNuX premium GUI, designed for compliance reviewers
Multi-user hosted dashboards / signed evidence portal None mapped cleanly to compliance JSON + tamper-proof ledger Build a portal yourself off the JSON outputs LeapNuX premium evidence portal
Per-auditor access controls DocuSign Rooms, ShareFile (generic) Add access control on top of your repo + cloud storage LeapNuX premium audit-room
A support contract with a real SLA None for open-source traceability tooling Hire a freelance compliance engineer LeapNuX premium support

Column A is what most teams reach for first — pair it with 5-NUX, no integration needed. Column B is the DIY path against the JSON output of every verb. Column C is when you want a turn-key commercial product with hosting, multi-user roles, and per-auditor access — that's LeapNuX premium.

For the full "what's enough" breakdown and comparison vs. enterprise tools — see docs/scope.md.


Where 5-NUX fits — the four-layer stack

5-NUX isn't a standalone product competing with anything. It's the project-management-and-audit layer in a four-layer stack you assemble from open-source pieces:

  ┌─────────────────────────────────────────────────────────┐
  │ LAYER 4 — YOUR PROJECT                                  │
  │   Whatever you're actually building — a fintech app,    │
  │   an e-commerce site, a health-tech platform, etc.      │
  └─────────────────────────────────────────────────────────┘
                              ▲
  ┌─────────────────────────────────────────────────────────┐
  │ LAYER 3 — WORKFLOW TOOLS  (third-party, open-source)    │
  │                                                         │
  │   • gstack — DEVELOPMENT workflow                       │
  │     /codex /cso /review /ship /qa /plan-eng-review      │
  │     "how to write, review, and ship code"               │
  │                                                         │
  │   • 5-NUX — PROJECT-MANAGEMENT + AUDIT workflow (this)  │
  │     rootnux trunknux branchnux leafnux fruitnux soilnux │
  │     "how to plan, verify, and audit what was built"     │
  │                                                         │
  │   Same audience, different jobs. They complement each   │
  │   other — they don't compete.                           │
  └─────────────────────────────────────────────────────────┘
                              ▲
  ┌─────────────────────────────────────────────────────────┐
  │ LAYER 2 — AI EXTENSION POINTS  (built into Claude Code) │
  │   Skills · Hooks · MCP · Sub-agents · Memory            │
  └─────────────────────────────────────────────────────────┘
                              ▲
  ┌─────────────────────────────────────────────────────────┐
  │ LAYER 1 — THE AI ASSISTANT ITSELF                       │
  │   Claude Code — Anthropic's AI agent in your terminal   │
  └─────────────────────────────────────────────────────────┘

One-liner: the project-management and audit layer for AI-assisted development. Pairs with gstack for development workflow. Both run on top of Claude Code.

Common questions

Question Answer
Isn't 5-NUX just gstack? No. gstack handles development workflow (review code, ship code, run QA). 5-NUX handles project management and audits (traceability, security reports, tamper-proof sign-off, risk register, decision records). Same audience, different jobs. A regulated team needs both.
Isn't this just Claude Code? No. Claude Code is the AI assistant. 5-NUX is the playbook that runs on top of it. Like asking "isn't Linux just the kernel?" — the kernel doesn't decide your build process.
Why not Jira / Linear / Notion / Confluence? Those are SaaS apps with their own database. 5-NUX lives in your git repo — every artifact is a markdown file, every command has a JSON output. Different category. Use them alongside 5-NUX (kanban in Linear, audit trail in 5-NUX).
Do I have to use Claude Code? No. The verbs run as plain Node.js — you don't need any AI assistant for the open-source path. Claude Code makes things faster; 5-NUX works with or without it.

Why a stack of small tools beats one big SaaS

  • No lock-in. Every layer is open-source and replaceable. Outgrow gstack? Swap it. Want to fork 5-NUX? Fork it. Your data is markdown in your git repo.
  • No procurement cycle. Three npm install -g commands instead of three vendor reviews.
  • Audit-ready by default. Your git history is the audit trail. The 5-NUX layer generates the reports auditors want — without any extra tooling.

The seven packages

The names follow a tree metaphor — each package maps to a project stage. The "just call it…" column is the friendly name we use in conversation.

Just call it… Package name What it does Status Commands
The Planning tool @leapnux/rootnux Specs, decision records, risks, knowledge base active init, lint, adr-new, risk-add, status, kb-init
The Sprint tool @leapnux/trunknux Sprint folders + weekly logs active new-sprint, summarize, lint, log
The Test tool @leapnux/branchnux Test plans, reports, AI-drafted planning active init, plan, codify, enrich, discover, batch-plan, report, validate, run, compare, visual baseline, visual compare, doctor, demo (~14 commands)
The Audit-Pack tool @leapnux/fruitnux Security reports, compliance JSON, sign-off, traceability active sca init, sca generate, sca pdf, sca oscal, sign, sign pdf, sign stale-check, br init, br link, br rtm, rtm (~11 commands)
The Health-Check tool @leapnux/leafnux Ongoing project-state snapshot active health
The Shared library @leapnux/6nux-core Internal — you don't run it directly active
The Bundle @leapnux/5nux One install that pulls in all five tools at once active

Install

Not yet on npm. The @leapnux/* namespace is being claimed; until then, clone this repo and run the package binaries from the workspace root.

# Coming soon (once the @leapnux namespace is claimed on npm):
npm install -g @leapnux/5nux        # the whole bundle (all five tools + shared library)
npm install -g @leapnux/rootnux     # just specs + decisions + risks + knowledge base
npm install -g @leapnux/trunknux    # just sprint folders
npm install -g @leapnux/branchnux   # just test planning + reports
npm install -g @leapnux/fruitnux    # just audit hand-off (security reports, compliance JSON, sign-off, traceability)
npm install -g @leapnux/leafnux     # just ongoing health check

Until then — install from source:

git clone https://github.com/leapnux/5nux.git
cd 5nux
npm install

Quick tour

Each line below is annotated with the friendly tool name in [brackets] so you can see which tool you're invoking.

# [Planning tool]
rootnux init                                          # set up REQUIREMENTS.md + traceability + risks + decision-record folders
rootnux adr-new "Use PostgreSQL for primary store"    # new decision record, auto-numbered
rootnux kb-init                                       # set up the knowledge-base scaffold (audit-prep sections)

# [Sprint tool]
trunknux new-sprint v1-launch                         # new dated sprint folder
trunknux summarize                                    # auto-generate a sprint summary from your git log

# [Test tool]
branchnux plan login                                  # AI-drafted test plan (every AI line flagged for human review)

# [Audit-Pack tool]
fruitnux rtm                                          # regenerate the traceability matrix
fruitnux sca init login                               # set up an 8-section security control report
fruitnux sca generate login                           # fill in the report from your test results
fruitnux sca oscal login                              # export it as government-standard compliance JSON
fruitnux sign login                                   # cryptographically sign the report — tampering is detectable

# [Health-Check tool]
leafnux health                                        # red / amber / green snapshot of project state

Full command reference: docs/reference.md. First-15-minutes walkthrough: docs/getting-started.md.


Who runs what — and where AI helps

Each tool serves a different role on the team, with a different relationship to AI-drafted content:

Tool Stage Who uses it What AI does here
Planning tool (rootnux) Requirements Product / project manager / spec author Drafts requirement entries, sets up decision records
Sprint tool (trunknux) Development Developer / engineer Summarizes the git log, drafts the sprint narrative
Test tool (branchnux) Test + verification QA / test lead Discovers test scenarios, drafts test plans (each AI line flagged for human review)
Health-Check tool (leafnux) Ongoing health Engineer / SRE Reads the artifacts, computes a red/amber/green status
Audit-Pack tool (fruitnux) Audit hand-off Compliance / legal / external auditor Generates security reports, compliance JSON, sign-off ledgers, traceability

For where the AI/human boundary sits at each stage, the four collaboration patterns (drafted-by-AI / confirmed-by-human, append-only enrichment, one agent per topic, cost-gated automation), and the typical end-to-end cycle — see docs/collaboration.md.


Folder layout in your project

The five tools produce peer folders at the root of your project — different roles, different folders, parallel views of the same project. Nothing is nested under another tool:

your-project/
├── requirements/                  ← Planning tool (rootnux) — Product / PM
│   ├── REQUIREMENTS.md            ← R-XX requirement entries
│   ├── TRACEABILITY.md            ← traceability matrix (generated by Audit-Pack tool)
│   ├── risks/risks.md             ← risk register
│   └── validations/<surface>/     ← test results + audit packets (Test + Audit-Pack tools)
│
├── docs/                          ← Planning tool (rootnux) — Product / PM
│   ├── adr/NNNN-<slug>.md         ← decision records (rootnux adr-new)
│   └── KNOWLEDGE_BASE.md          ← knowledge-base scaffold (rootnux kb-init)
│
├── sprint-log/                    ← Sprint tool (trunknux) — Dev / Eng
│   └── <date>_<slug>/
│       ├── README.md              ← sprint scaffold (trunknux new-sprint)
│       ├── LOG.md                 ← weekly journal entries (trunknux log)
│       └── SPRINT_SUMMARY.md      ← git-log roll-up (trunknux summarize)
│
├── testing-log/                   ← Test tool (branchnux) — QA / test lead
│   └── <date>_<surface>/
│       ├── test-plan.md           ← test cases + Given/When/Then steps (branchnux plan)
│       ├── execution-log.md       ← run results (auto-generated)
│       └── evidence/              ← screenshots, logs (auto)
│
├── <your-app-source>/             ← your code (whatever framework — outside 5-NUX scope)
│
└── (Health-Check output: lives in your CI / dependabot / observability tools — not in the repo)

The five tools do not import each other — they coordinate through file conventions in @leapnux/6nux-core. Implementation details: docs/ARCHITECTURE.md.


📚 Documentation

  • docs/MOTTO.md — what's free vs. what's a paid premium offering (and why we keep that line sharp)
  • docs/scope.md — what 5-NUX is, what it isn't, what's enough on its own; comparison vs. DOORS / Polarion / Jama
  • docs/collaboration.md — who does what, and how AI and humans collaborate at each stage
  • docs/6-NUX.md — the conceptual model (root / trunk / branch / leaf / fruit / soil)
  • docs/ARCHITECTURE.md — implementation details (monorepo layout, package contracts, security model)
  • docs/concepts.md — three-track discipline, [VERIFY] markers, the cryptographic chain, deterministic-core vs. opt-in-AI split
  • docs/getting-started.md — your first 15 minutes
  • docs/reference.md — full command reference
  • docs/adr/ — decision records (7 locked architectural decisions, all written via rootnux adr-new itself)
  • CHANGELOG.md — release history
  • CONTRIBUTING.md — how to contribute (commit sign-off, hygiene rules, workspace workflow)

Roadmap

  • v0.4.x — alpha series; rootnux + trunknux + branchnux mature; leafnux + fruitnux brought into active scope
  • v0.5.0-alpha.1trunknux log, rootnux kb-init, leafnux health shipped; fruitnux pack in design
  • v0.6 — direct Claude Code integration (Anthropic's Model Context Protocol — wires 5-NUX commands straight into your AI assistant via .claude/settings.json)
  • v1.0 — stability + landing page at leapnux.com + the commercial premium spec

Open an issue to vote on what to prioritize next.


Contributing

See CONTRIBUTING.md. Quick version: sign your commits with git commit -s (a one-line "I wrote this" certificate — no extra paperwork to sign), include tests with every pull request, open an issue first for anything large.


License

Apache 2.0. See LICENSE.

"BranchNuX™" and "LeapNuX™" are trademarks of Chu Ling. See NOTICE for trademark terms. The Apache 2.0 license covers the code; the trademark covers the name.

Author

Chu Ling — [email protected]. Security reports: GitHub Private Vulnerability Reporting (preferred).

About

AI-native OSS PM tool in CLI. Purpose-built for regulated software — auto-generates RTM, SCA, OSCAL, HMAC-signed audit packages from plain files in your repo. SOC 2 / ISO 27001 / NYDFS / GDPR ready. Free.

Topics

Resources

License

Contributing

Security policy

Stars

Watchers

Forks

Packages

 
 
 

Contributors