ADLC plugin
ADLC is a set of dependency-free CLIs and agent plugins: gates at each development phase against typical model failure modes.
- 54 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The installer runs a script from the network and changes the settings of detected agents.
- Gates read the repository and write tickets and evidence into `.adlc/`.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/adlcDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The installer runs a script from the network and changes the settings of detected agents.
- Gates read the repository and write tickets and evidence into `.adlc/`.
This is third-party code. Review the repository files before installing.
What it does
ADLC is the Agentic Development Lifecycle Toolkit: a workspace of small Node.js tools, each checking one condition during agent-driven development. The author builds the lifecycle around model failure modes: premature satisfaction, sycophancy, context rot, confident hallucination and reward hacking, rather than human weaknesses. The tools are grouped: spec and ticket shaping (`spec-lint`, `premortem`), execution supervision (`preflight`, `model-router`), review evidence (`behavior-diff`, `hollow-test`) and compounding lessons (`lesson-foundry`, `skill-rot`). `adlc init` bootstraps the `.adlc/` directory in a repository. The installer wires integrations for Claude Code, Codex, Copilot, pi, Antigravity and herdr.
Who it is for. Teams that run most development through agents and want mechanical checks instead of trusting a model's word.
Good fit when
- A spec should be linted before it goes to an agent.
- An agent's test looks suspiciously hollow and needs a hollowness check.
- Behavior before and after a change must be compared as evidence for review.
- Lessons from failures should accumulate in the repository rather than get lost.
Not a fit when
- The project is small and gates are excessive.
- Windows support without WSL is needed.
- A simple skill is wanted instead of a toolkit with its own doctrine.
Example request
Initialize ADLC in the repository and run `adlc spec-lint` on the new feature spec.Limitations
macOS or Linux and Node.js 18 or newer are required. On Windows many checks break on absolute paths, so use WSL. The installer runs as a script from the network, so read it before running. For Cursor, plugins install through the in-app marketplace, OpenCode must be run inside a repository, and pi needs Node.js 22.19 or newer. The installer skips agents that are not on the machine.
How to disable. Remove the integrations from agent settings and the toolkit with the uninstall steps from its install directory. Delete the `.adlc/` directory with tickets and evidence in repositories by hand.
FAQ
How do I install it?
Run `curl -fsSL https://www.agenticlifecycle.ai/install.sh | sh`, then `adlc init` in a repository. The script finds agents on the machine.
Which agents connect automatically?
Claude Code, Codex, GitHub Copilot, pi, Google Antigravity and herdr. Cursor installs through the in-app marketplace, and OpenCode is set up inside a repository.
How does it differ from ordinary SDLC?
Each phase and gate traces to a specific model failure mode such as sycophancy or context rot. Phases built for human weaknesses are cut.
Are there dependencies?
No, each tool is written without dependencies and the shared runtime uses the `.adlc/` directory.
Does it work on Windows?
No, use WSL: many gates fail on absolute Windows paths.
Discussion
No questions or reviews yet.
Related
PluginsADLC Harness takes a task from intent to PR with one command, and the human approves only four key gates.
Skills that give a coding agent your team rules at session start: a rule index, a task factory, decision records and evals
The AI-SDLC plugin for Claude Code carries a task from spec to PR: a readiness gate, a developer agent and several reviewers.
Nine plugins for Claude Code with 51 skills and 2 agents: a single workflow from a rough idea to a reviewed pull request.
A set of agents, skills and review workflows for AL development in Microsoft Dynamics 365 Business Central: specs, tests, checks.
Skill and plugin for Codex and Claude Code: the agent drives a project through a plan-to-maintain loop with human gates.
README in short
Each tool checks one condition and leaves evidence in `.adlc/`, so the result reads as a machine record rather than a model's argument.