Stop duplicating agent skills across repos
It centralizes reusable coding-agent workflows so teams can link to canonical playbooks instead of copying long SKILL.md files.

What it does
dotai is a shared library of agent skills—operational playbooks for coding agents. Each skill defines a routing workflow, from test-driven development loops to root-cause debugging and delete-first repo cleanup. Repositories reference these canonical definitions instead of maintaining their own lengthy instruction files.
The interesting bit
Skills serve as the main routing layer for agents, not just static prompts. The repo explicitly models dependencies between skills—debug optionally pulls in tdd, autogoal optionally activates orchestrator—which is more structure than a typical folder of markdown snippets.
Key highlights
- Six built-in skills:
autogoal,debug,hard-cut,orchestrator,tdd, andvideo-transcripts. - Dependency chains between skills (e.g.,
debugoptionally pulls intdd). scripts/validate-skillsfor checking edits.- Strict editing rules: operational bodies, no secrets, no private hostnames.
Caveats
- Hard dependencies on Codex-specific APIs and external binaries (
curl,jq, Gemini keys) for several skills. - The
skillsCLI and its versioning mechanism are not documented in the README.
Verdict Teams running agent-driven development across multiple repos get the most value; solo developers maintaining a single codebase probably do not need the abstraction.
Frequently asked
- What is udecode/dotai?
- It centralizes reusable coding-agent workflows so teams can link to canonical playbooks instead of copying long SKILL.md files.
- Is dotai open source?
- Yes — udecode/dotai is an open-source project tracked on heatdrop.
- What language is dotai written in?
- udecode/dotai is primarily written in JavaScript.
- How popular is dotai?
- udecode/dotai has 1.2k stars on GitHub.
- Where can I find dotai?
- udecode/dotai is on GitHub at https://github.com/udecode/dotai.