antonio-mastropaolo/dsh-model-hub

C2

DeepSeek Harness plugin: model hub. DSH-native sign-in (Claude, Codex, Copilot, OpenRouter, xAI, Kimi), rocco GPU dispatch, per-model cost, spend caps, prompt-cache checks, and the dsh-hub CLI.

★ 1+ · antonio-mastropaolo/dsh-model-hub source on GitHub · this plugin in the registry

antonio-mastropaolo/dsh-model-hub is a DeepSeek Harness plugin rated C2 — one powerful capability or sensitive behavior. It gates tool execution, command execution in build output only, network server in build output only, reads credential-class env vars.

Not installable — declares only dsh.client, which dsh plugin add cannot install

What it can do

CapabilityFlagEvidence
gates tool executiontool_gatehook: tools/pre-execute
command execution in build output onlyexec_bundled×3 in build output only, e.g. lib/dsh-messages.mjs:4, lib/dsh-messages.mjs:4
network server in build output onlynet_server_bundled×1 in build output only, e.g. lib/index.js:147
reads credential-class env varstoken_envANTHROPIC_API_KEY, ANTHROPIC_AUTH_TOKEN

Tool registrations

0 static tool registration point(s).

Services it injects

sessions slots

Hooks it attaches

llm/stream session/event tools/pre-execute

Outbound domains

api.anthropic.com ollama.com

Environment variables it reads

ANTHROPIC_API_KEY ANTHROPIC_AUTH_TOKEN DSH_HOME DSH_HUB_REFRESH

How to read this

Levels measure capability surface and transparency, not maliciousness. A C3 plugin can be entirely legitimate — a desktop shell genuinely needs subprocesses. The point is that you can see this before installing. See the levels explained and how dsh plugins work.

Findings come from static analysis of shipped code; nothing is executed. Think a flag is wrong? Open an issue — every flag cites the file and line it came from.