Mockingbird — AI skill for Claude Code
Pinned Claude Code with isolated CLIProxyAPI custom-model routing.
How to install Mockingbird
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open DigiBugCat/mockingbird and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Mockingbird does
Pinned Claude Code with isolated CLIProxyAPI custom-model routing.
Alternatives in AI
- Zen MCP Server — The power of Claude Code + Gemini / OpenAI / Grok / OpenRouter / Ollama / Custom Model working as one 11.3k ★
- Oh My Claudecode — 32 specialized agents and 7 execution modes for Claude Code, with smart model routing and automatic paralleliz 10.9k ★
- Cc Mirror — Create multiple isolated Claude Code variants with custom providers (Z.ai, MiniMax, OpenRouter, LiteLLM) 2.1k ★
README
Mockingbird
Mockingbird freezes Claude Code at 2.1.259 and adds a local, isolated custom model path backed by CLIProxyAPI. Native Claude models, Anthropic login, claude.ai connectors, and account features continue to use Claude Code's unchanged first-party transport.
[!WARNING] **Use at your own risk.** Mockingbird runs a modified, unsupported third-party client. It may break without notice, incur provider charges, or affect account access. Treat patched use as observable: the client retains Claude Code's [operational telemetry and error-reporting paths](https://code.claude.com/docs/en/data-usage) and sends normal [client/version metadata](https://code.claude.com/docs/en/llm-gateway-protocol), while provider logs, traffic patterns, or future integrity checks may also reveal modifications. Mockingbird does not attempt to hide changes or guarantee undetectability.
The custom picker contains:
| Model | Best for | Effort | Default | Tier |
|---|---|---|---|---|
gpt-5.6-luna |
Bounded, repeatable, high-volume work | none, low, medium, high, xhigh, max | medium | priority |
gpt-5.6-sol |
Complex, ambiguous, or polished high-value work | none, low, medium, high, xhigh, max | medium | priority |
grok-4.5 |
Fast multi-step engineering and agentic tool use | none, low, medium, high | high | provider default |
`/model` selects the main-session model. Agent model fields accept any of the same IDs. Built-in Explore is pinned to GPT-5.6 Luna at `max`; Terra is excluded from the catalog and priority policy. GPT priority service is applied independently of the selected effort, including internal compaction traffic. Ultracode remains Claude Code's xhigh-plus-delegation workflow.
The patched CLI adds a neutral model-routing ladder to the same shared environment guidance used by the main session and subagents. It describes Claude, GPT-5.6, and Grok by workload without changing any model's personality.
Native Claude routing includes Clau
Related Skills
HomeAILab
Home AI inference lab running Qwen3.8 across RTX 5090, RTX 3090, and DGX Spark with vLLM, llama.cpp, custom Go
MCP Memory Agent
A four layered memory system for LLM agents : working, semantic, episodic memory and a routing layer exposed a
CLIProxyAPI
Wrap Gemini CLI, Antigravity, ChatGPT Codex, Claude Code, Qwen Code, iFlow as an OpenAI/Gemini/Claude/Codex co
Agent Plane
A multi-tenant platform for running AI agents in isolated Vercel Sandboxes, exposed via a REST API. Supports a
Harness Arena
Blind, LMSYS-style arena for comparing AI coding agent harnesses (Claude Code, Codex CLI, OpenClaw, Hermes, On
Harness Arena Fe
Blind, LMSYS-style arena for comparing AI coding agent harnesses (Claude Code, Codex CLI, OpenClaw, Hermes, On
Related Agents
Lending Specialist
Lending-protocol specialist. Aave V3, Compound V3, Morpho, Silo, Euler, custom lending. Use when the target is
Pocketpal Intake
Intake and routing stage for PocketPal development tasks. Creates isolated worktree, parses issues/tickets, cl
Miner High
Delegate worker pinned to HIGH reasoning effort, the Opus 4.8 model default. Pick for diagnosis, multi-constra