AI coding agents can generate code. AIDDbot builds software you can trust.

Why you need AIDD

Agents are strong on isolated tasks. On real projects, three failure modes show up repeatedly.

//Detail or invent

Missing context — or plausible fabrication. Spec-driven development keeps intent in the repo.

//Guide or chaos

Code that ignores your standards. Rules over tools — versioned with the project.

//Verify or hope

Silent drift until fixes are expensive. Human in the loop at the one checkpoint that matters.

Three doors

Four commands cover the lifecycle. Each chains skills in fresh subagents so nothing inherits the previous step's clutter.

//explore-and-extract

Document what is there — architecture, schemas, coding rules. Works on empty and mature repos alike.

//spec-feature

Turn a requirement into a spec, then plan, build, verify, and release.

//explore-and-refactor

Re-explore for technical drift, pick defects, and drive each fix through a refactor spec.

Plain markdown. Any agent.

No CLI, no package install — copy `.agents/` and run. Same skills in the editors you already use.

//Antigravity · Cursor · VSCode · JetBrains

IDEs that load project context from your repo.

//Claude Code · Codex · Copilot · OpenCode

Agent harnesses that read or invoke skills.

Open source, production-minded

AIDDbot is an open-source initiative by Alberto Basalo. Clone the skills, ship with confidence.