pi-plugin-cc — drive the Pi coding agent from Claude Code 🥧

English · 中文

External references: Pi coding agent · Pi RPC mode · Pi providers

A Claude Code plugin that delegates reviews and coding tasks to the Pi coding agent. 1:1 fork of codex-plugin-cc, runtime swapped from Codex to Pi.

The hard dependency is pi, not any particular LLM. Pi can be configured for DeepSeek, OpenAI, Anthropic, Google, Ollama, LM Studio, or any OpenAI-compatible endpoint via ~/.pi/agent/models.json. The plugin defers all model selection to pi unless you override per command.

• Code review against the working tree or a branch base, with structured findings
• Adversarial review that challenges the design — not just spell-checks the diff
• Task delegation for diagnoses, refactors, and longer rescues, foreground or background
• Background job control — status, result, cancel, and stop-time review gate
• No OAuth — pi authenticates by API key (provider-specific), no codex login required

Plays well with pi-subagents: if installed, it works inside /pi:rescue runs without configuration.

🔄 How it works

Codex's broker layer is gone — Pi is one-conversation-per-process, so the plugin spawns a fresh pi --mode rpc for each task. Background jobs are tracked in workspace-scoped state files. Review prompts inline the JSON schema since Pi has no outputSchema knob.

Slash commands

Every command accepts --model <id> to pin a specific model just for that run. With no --model and no env override (see Pick your model), pi falls back to whatever it has configured by default.

Quick Start

# 1. Install pi (required)
npm install -g --ignore-scripts @earendil-works/pi-coding-agent

# 2. Configure a provider — pick one
export OPENAI_API_KEY=sk-...           # OpenAI
export ANTHROPIC_API_KEY=sk-ant-...    # Anthropic
export GOOGLE_API_KEY=...              # Google
export DEEPSEEK_API_KEY=sk-...         # DeepSeek
# or run a local model: see https://github.com/earendil-works/pi (Ollama / LM Studio)

# 3. Verify pi sees a model
pi --list-models | head

Install the plugin in Claude Code:

> /plugin marketplace add Agents365-ai/pi-plugin-cc
> /plugin install pi@agents365-pi
> /reload-plugins
> /pi:setup

/pi:setup returns a readiness report. If pi is missing and npm is on PATH, it offers to install it for you.

Usage

> /pi:review
> /pi:review --base main
> /pi:review --model claude-sonnet-4
> /pi:adversarial-review focus on the new auth middleware
> /pi:rescue investigate why the Windows CI build is failing
> /pi:rescue --background --model gpt-4o refactor src/payments/
> /pi:status
> /pi:status task-mpgyiwb9-e3k641 --wait
> /pi:result task-mpgyiwb9-e3k641
> /pi:cancel task-mpgyiwb9-e3k641

--effort <off|minimal|low|medium|high|xhigh> is passed through to Pi via set_thinking_level. Models that do not support thinking silently ignore it (the plugin logs a one-line note to stderr when this happens).

Pick your model

The plugin keeps three layers of model resolution:

Layer 1 wins over layer 2 wins over layer 3. None of the layers are required — leave them all unset and pi picks for you.

Suggested settings by provider

These are opinions, not requirements. Pick what fits your latency / cost / quality budget.

To make these defaults sticky:

export PI_PLUGIN_REVIEW_MODEL=claude-haiku-4-5
export PI_PLUGIN_ADVERSARIAL_REVIEW_MODEL=claude-sonnet-4-6

Configure pi

The minimum is a single env var with your API key. For richer setups, write ~/.pi/agent/models.json:

{
  "providers": {
    "deepseek": { "apiKey": "sk-..." },
    "openai":   { "apiKey": "sk-..." },
    "anthropic":{ "apiKey": "sk-ant-..." },
    "google":   { "apiKey": "..." },
    "openrouter": {
      "api": "openai-completions",
      "apiKey": "sk-or-v1-...",
      "baseUrl": "https://openrouter.ai/api/v1",
      "models": [
        {
          "id": "deepseek/deepseek-chat",
          "name": "DeepSeek via OpenRouter",
          "contextWindow": 128000,
          "maxTokens": 8192,
          "input": ["text"]
        }
      ]
    }
  }
}

Built-in providers (anthropic, openai, google, deepseek, ollama, lmstudio) only need apiKey (and optional baseUrl); pi ships their model lists. For custom OpenAI-compatible endpoints, set api: "openai-completions" and declare the models you want exposed.

Full reference: pi providers docs.

Stop-time review gate

Opt in with /pi:setup --enable-review-gate. When a Claude session ends, the plugin runs a Pi adversarial review of the previous turn and can block the stop if it finds material issues. If pi is unavailable while the gate is enabled, the hook blocks (it does not silently let the session end). Disable with /pi:setup --disable-review-gate.

🔗 Related projects

❤️ Support

If this plugin helps you, consider supporting the author:

WeChat Pay

Alipay

Buy Me a Coffee

Give a Reward

👤 Author

Agents365-ai

• GitHub: https://github.com/Agents365-ai
• Bilibili: https://space.bilibili.com/441831884

📄 License

Apache License 2.0. Forked from codex-plugin-cc (Apache-2.0, OpenAI) — see NOTICE.