Octohub — Data skill for Claude Code

OctoHub

One OpenAI-style API in front of all your LLM providers — with multi-tenant keys, full request logging, and usage analytics.

[](https://github.com/Muvon/octohub/actions/workflows/ci.yml) [](https://github.com/Muvon/octohub/releases) [](LICENSE)

OctoHub is a high-performance LLM proxy written in Rust. Clients talk to one stable endpoint; OctoHub talks to whichever providers you configure — OpenAI, Anthropic, Google, DeepSeek, Ollama, OpenRouter, Modal-hosted vLLM, and more — and records every request and response in your own database.

            ┌──────────────────┐
 client  →  │  OctoHub proxy   │  →  provider A (openai)
            │  (Rust / hyper)  │  →  provider B (anthropic)
            │                  │  →  provider C (ollama, your GPU)
            └──────────────────┘  →  provider D (Modal-hosted vLLM)
                     │
                     └─→ your DB (SQLite / MySQL / PostgreSQL)

Why OctoHub?

Provider APIs don't give you per-customer usage tracking, audit logs, or a way to swap models without breaking clients. OctoHub is the missing front door: one interface, many backends, everything recorded.

Quick Start

# Build
cargo build --release

# Drop in a config
cat > octohub.toml <<'EOF'
[server]
host = "127.0.0.1"
port = 8080
api_key = "your-master-secret"
db_url = "sqlite://octohub.db"

[models]
"minimax-m2.7" = ["minimax:minimax-m2.7"]
EOF

# Run
./target/release/octohub

Create a client key and make your first call:

curl -sX POST http://127.0.0.1:8080/v1/admin/keys \
  -H "Authorization: Bearer your-master-secret" \
  -H "Content-Type: application/json" \
  -d '{"name": "my-app"}'
# → {"id":1,"key":"nYwT8kQ2v…", ...}

curl -sX POST http://127.0.0.1:8080/v1/completions \
  -H "Authorization: Bearer