Disciplined Agentic Engineering plugin
A set of three Claude Code plugins: specs, acceptance tests, mutation testing and change-risk analysis.
- 163 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The plugins run commands and tests in the project.
- The agent writes spec files and code under the checks.
Install
- Works in
Manual install
/plugin marketplace add swingerman/engineerThen `/plugin install engineer@disciplined-agentic-engineering`; `atdd` and `crap-analyzer` install from the same marketplace.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The plugins run commands and tests in the project.
- The agent writes spec files and code under the checks.
This is third-party code. Review the repository files before installing.
What it does
Disciplined Agentic Engineering, or DAE, guides agent development along a pipeline: feature description, acceptance criteria, a Gherkin spec and an architecture plan. Each layer is approved before the next is written. Discipline rests on deterministic tools the agent must satisfy. The kit has three plugins: `engineer` with the `/engineer.onboard`, `/engineer.next` and `/engineer.discuss` commands, `atdd` with acceptance tests and differential mutation testing and `crap-analyzer` for change-risk analysis. The methodology is host-independent and the reference host is Claude Code. A handoff log lives in the feature folder.
Who it is for. Engineers who hand coding to an agent but keep architecture, behavior contracts and verification.
Good fit when
- A new feature goes from idea to spec with approval at each step.
- You need acceptance tests and a mutation check that tests catch changes.
- Change risk is assessed with `crap-analyzer` before merge.
- At session start `/engineer.next` reminds you what to do next.
Not a fit when
- You need a quick prototype with no specs: the pipeline slows work down.
- The project is not on Claude Code: the plugins target it.
- The team will not approve each document layer.
Example request
Run /engineer.onboard for the project and describe the first feature: goal, acceptance criteria and spec.Limitations
The process is heavy: it adds feature folders, a manifest, a handoff log and approvals that pay off on large tasks. Today the plugins run in Claude Code and porting to another host is described in the Host independence section. The repository was renamed from `atdd` and `disciplined-agentic-engineering`, while the marketplace kept its former name. Mutation testing lengthens a run noticeably: on large projects run it on the diff and its changed files.
How to disable. Run `/plugin uninstall` for each of the three plugins; delete feature folders in the project by hand.
FAQ
Which plugins are included?
`engineer` with the methodology, `atdd` with acceptance tests and mutation testing and `crap-analyzer` for change-risk analysis. They install separately.
How do I run it from a clone?
Clone the repository and run `claude --plugin-dir ./engineer`.
What does a feature folder hold?
The `feature.md`, `acs.md`, `spec.md` and `plan.md` files, the `handoffs/` log and generated pipelines in `.build/`.
Does it work in Codex or Cursor?
The methodology is host-independent but ready plugins are shipped for Claude Code today.
Why mutation testing?
It mutates code and checks that tests fail. That shows acceptance tests actually catch breakage.
Discussion
No questions or reviews yet.
Related
PluginsA plugin with three skills for large development work: project analysis, phased tasks tracked as GitHub Issues, parallel agents and code review
Claude Code plugin with 17 skills: task sizing, design review, TDD, fresh-context subagents and a PR after three human decisions.
A Claude Code plugin for development with separated roles: spec, review, TDD, QA and verification by different agents
ADLC is a set of dependency-free CLIs and agent plugins: gates at each development phase against typical model failure modes.
The AI-SDLC plugin for Claude Code carries a task from spec to PR: a readiness gate, a developer agent and several reviewers.
Skills for Claude Code and other agents: spec pushback, six-specialist branch review, architecture and accessibility audits
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Disciplined Agentic Engineering (this card) | no | 163 | |
| PAUL | no | 1,269 | PAUL runs a plan, execute and close loop; DAE adds Gherkin specs and mutation testing |
| ProductSpec | no | 317 | ProductSpec defines a product spec format; DAE builds a pipeline from feature to plan and tests |
README in short
The repository sets out an engineering methodology for AI development: specs, tests and guardrails instead of prompt rules. It has a quick start, a pipeline description, tunable autonomy, three plugins, a host independence section and arguments against vibe coding.