dragoman-ai 0.6.0

A small CLI that lets Claude Code reach non-Anthropic models — Ollama, Perplexity, OpenAI, Gemini — through a subagent that knows how to call it.

dragoman

A small CLI that lets Claude Code reach non-Anthropic models — Ollama (local), Perplexity (search-augmented), OpenAI, Gemini, anything OpenAI-compatible — through one verb the existing subagent runtime can call.

I have a GPU running Ollama. I pay for OpenAI, Gemini, and Perplexity because each is the right answer for a different shape of question. Claude Code is the conductor. Dragoman is the verb that lets the conductor talk to the rest of the orchestra.

v0.6.0 alpha. Apache 2.0.

What it does

Two commands. No agent loop. No tool execution. No shell. Just one HTTPS call per ask.

dragoman ask --model perplexity:sonar-pro --prompt "..."   # one HTTPS call, prints text
dragoman models                                            # what's configured, one per line

The persona injected by dragoman init teaches Claude Code when to spawn a real Task() subagent that uses dragoman ask for the cognitive step. All filesystem and shell work happens through Claude Code's normal tools — the harness's audit, fan-out, and permissions stay intact. Dragoman holds keys; the harness holds the runtime.

Install

You can install Dragoman securely using any modern Python tool manager:

# Using pipx (Standard)
pipx install dragoman-ai

# Using uv (Fastest)
uv tool install dragoman-ai

(Homebrew support is currently being built!)

Then, initialize your config and provider keys:

dragoman init

Open a fresh Claude Code session. Try: "What's the best model for [your task]?" If the dragon shows up, it works.

Keys live where you already keep them

API keys can be literal strings, environment variables, or references resolved at call time:

[providers.perplexity_1]
type = "openai_compat"
host = "https://api.perplexity.ai"
api_key = "op://Personal/Perplexity/credential"   # 1Password CLI

[providers.groq_1]
type = "openai_compat"
host = "https://api.groq.com/v1"
api_key = "keychain://groq/apikey"              # macOS Keychain

Dragoman fetches by reference, uses the key for one HTTPS call, discards it. The key never enters Claude's context.

Multiple providers, multiple accounts

Dragoman replaces hardcoded singleton endpoints with a dynamic provider registry. You can connect as many distinct accounts, gateways, or local instances as you want simultaneously.

For example, if you run Ollama on a laptop and on a separate workstation reachable via Tailscale:

[providers.laptop_1]
type = "openai_compat"
host = "http://localhost:11434/v1"

[providers.workstation_1]
type = "openai_compat"
host = "http://workstation.tailnet.ts.net:11434/v1"

Then you simply tell Claude exactly which pipe to use: --model workstation_1:qwen2.5:72b. No magic network probing; just explicit, unopinionated routing.

What it writes to your system

Artifact	Path	Created by	Removed by
Provider config	~/.config/dragoman/config.toml	dragoman init	dragoman uninstall --purge-config
Persona block	~/.claude/CLAUDE.md (or project)	dragoman init	dragoman uninstall
Python package	active env	pip install dragoman-ai	pip uninstall dragoman-ai

The persona block is bracketed by <!-- dragoman persona ... --> markers; removal is exact and idempotent.

Telemetry

None. Dragoman makes no outbound calls of its own — only to provider endpoints you configured.

Contributing

See CONTRIBUTING.md. Small PRs welcome; commits need DCO sign-off (git commit -s); Contributor Covenant applies.

License

Apache 2.0. See LICENSE.

---

Dragoman was the translator-fixer at Ottoman, Levantine, and European courts. The English word and the Hebrew meturgeman share an Akkadian root and have nothing to do with reptiles. The 🐉 emoji is a typo I refuse to fix. There is also no evidence that dragons were actually reptiles.