$0

Clancy — your coding agent on dLLM

Clancy is a full-featured terminal coding agent — backed by the dLLM network: private, uncensored, decentralized. It runs OpenCode (a mature agent with MCP, subagents, sessions, compaction, LSP and a web UI) skinned and wired to dLLM, so you get best-in-class tooling on our models and network instead of one company's servers.

Install

One line. Installs OpenCode if needed, drops in the Clancy config, and creates the clancy command:

curl -fsSL https://dllmnetwork.com/clancy/install.sh | sh

Open a new shell (or source ~/.bashrc), then just type clancy.

Get your key

Clancy uses your dLLM account key — the same anonymous key that holds your credit. Grab it from your Account page, then save it once:

echo YOUR_KEY > ~/.clancy/key

No balance yet? Add credit — pay-as-you-go in crypto, no subscription. You only spend per token, on whichever model you pick.

Use it

clancy # open the TUI in your project directory clancy run "fix the failing test" # one-shot, non-interactive /models # switch model inside the TUI

The default is GLM 5.2 — a frontier open-weight model on our Anonymous tier (not tied to your identity, processed by a hosted provider; not sealed). Want your prompts sealed end-to-end? Switch with /models to our uncensored E2EE worker — those we genuinely can't read.

The models

GLM 5.2
frontier · default
Qwen3.6 27B
uncensored · E2EE worker
Kimi K2.7 Code
code-specialized
Claude Sonnet / Opus
top-tier coding
DeepSeek · Qwen
fast & cheap

Every model is billed by the token at the live network price — the Pricing page lists them all.

What's ours vs. OpenCode

OpenCode is the engine (open source). Clancy is the dLLM provider config, the theme, and the launcher on top — no fork, nothing to lock you in. The differentiation is the network: private, uncensored, real model choice, and the option to run on your own infrastructure.

Your account & key →  ·  Add credit →  ·  Models & pricing →