Codegeist AI is an open-source project family building practical AI systems and
the reproducible tooling behind them. Codegeist
is the coding agent. Codegeist OS
and Codegeist LLM explore a
more local, inspectable, and deliberately separated AI stack.
The projects are at different stages. The coding agent is usable early-stage software; Codegeist LLM publishes narrow research artifacts; Codegeist OS is currently architecture and research, not a released operating system.
| Goal | Destination |
|---|---|
| Use or contribute to the coding agent | codegeist |
| Explore the operating-system architecture | codegeist-os |
| Follow local-model research and artifacts | codegeist-llm and Hugging Face |
| Follow current priorities | Public roadmap |
| Read the contribution workflow | Contributing guide |
| Ask for quick help | Discord developer community |
| Project | Current role | Status |
|---|---|---|
codegeist |
Open-source Java coding agent with CLI/TUI workflows, tools, MCP, local and hosted model support, and native releases | Early-stage software |
codegeist-os |
AI-focused operating-system architecture built around explicit identities, policy, isolation, and recovery | Architecture and research |
codegeist-llm |
Compact local-model research, provenance, evaluation, and reproducible artifact work | Early research and artifacts |
These projects do not yet form a completed integrated stack. Their repositories separate current evidence from planned architecture and publish progress in the open.
codegeist-agent-kit- shared OpenCode rules, commands, skills, integrations, and generated workspace releases.codegeist-devcontainer-kit- reproducible VS Code devcontainer foundations and generated consumer releases.
- Source, project planning, Issues, releases, and durable technical decisions live on GitHub.
- Public model profiles and reviewed model artifacts are discoverable through
huggingface.co/codegeist. - Discord supports quick help, feedback, and workflow sharing; decisions that need to last move back into repositories and documentation.
- Claims should identify whether they describe released software, an experiment, or intended architecture.
- Choose the repository that owns the change.
- Read its local guidance and find or open a focused Issue.
- Work in a small branch or fork and run the documented check.
- Open a pull request that links the Issue and any applicable local task.
The account-wide contribution guide applies unless a repository provides more specific instructions. General support routes are documented in the support guide, and vulnerabilities belong in the security process.
Canonical repository checks
| Repository | Check |
|---|---|
codegeist |
task cli:check |
codegeist-agent-kit |
task test |
codegeist-devcontainer-kit |
task check |
.github |
bash scripts/check.sh |
codegeist-ai/codegeist-ai |
bash scripts/check.sh |
codegeist-ai is currently a personal GitHub account used only as the public
Codegeist AI brand account. It is not a GitHub organization. The
codegeist-ai/codegeist-ai
repository renders this profile, while
.github owns account-wide community
defaults. The complete ownership and release model is documented in
GITHUB_ACCOUNT_MODEL.md.
Codegeist-owned profile material is available under the 0BSD license.