Build, run, and scale autonomous AI CTO workflows across coding, inference, and multi-agent operations.
HeySalad CTO is a unified control repo for three execution layers:
cheri-ml: inference and model-serving runtimesheri-ml: Codex-style coding workflowsopencto: multi-agent orchestration and operations
The goal is simple: one operating system for AI engineering teams.
# 1) Inference server
cd cheri-ml
python serve_model.py
# 2) Codex workflows
cd ../sheri-ml/codex-rs
cargo build --release
# 3) Agent orchestration
cd ../../opencto/Sheri-ML/sheri-ml-cli
npm install
npm startCTO/
├── cheri-ml/ # inference server
├── sheri-ml/ # coding/runtime tooling
├── opencto/ # agent orchestration
├── docs/ # architecture + guides
├── AGENTS.md
├── CLAUDE.md
└── VISION.md
This project is open source under Apache-2.0 and intended for broad adoption.
If your team wants business implementation support (deployment, integration, managed operations), see BUSINESS.md.
- License: LICENSE (Apache-2.0)
- Attribution: NOTICE and ATTRIBUTION.md
- Trademark policy: TRADEMARKS.md
- HeySalad is a registered trademark (UK00004063403)
- Website: https://heysalad.io
- Business implementation: investors@heysalad.io
- Company: HeySalad Inc., 584 Castro St, Suite #4003, San Francisco, CA 94114, US
- Pull requests to
mainrun quality checks via.github/workflows/pr-quality.yml. - Security and smoke checks run via
.github/workflows/ci-security.yml. - Releases are created from tags like
v1.2.3(or manual dispatch) via.github/workflows/release.yml.
Example release flow:
git checkout main
git pull
git tag v1.0.0
git push origin v1.0.0- OpenCTO docs index: docs/opencto/README.md
- Platform spec: docs/opencto/OPENCTO_PLATFORM_SPEC.md
- Frontend, brand and monetisation spec: docs/opencto/OPENCTO_FRONTEND_BRAND_MONETISATION_SPEC.md
- Implementation roadmap: docs/opencto/IMPLEMENTATION_ROADMAP.md