Pi

Pi is a terminal coding agent. Install the CLI from npm:

npm install -g --ignore-scripts @earendil-works/pi-coding-agent

Base Invocation

The built-in Dagu provider invokes Pi in print mode:

pi -p "<prompt>"

Common Flags

Flag	Type	Description
--provider	string	LLM provider, such as openrouter, openai, or anthropic
--model	string	Model pattern or ID, including provider/model and optional :<thinking>
--thinking	string	Thinking level: off, minimal, low, medium, high, or xhigh
--tools	string	Comma-separated tool allowlist
--exclude-tools	string	Comma-separated tool denylist
--no-tools	boolean	Disable all tools
--no-session	boolean	Run without saving a session
--no-context-files	boolean	Disable AGENTS.md and CLAUDE.md discovery
--list-models	string	List available models, optionally filtered

Example

Use a custom harness when a Pi option needs an exact flag mapping. Dagu uses with.provider to select the harness provider, so this example maps ai_provider to Pi's --provider flag.

harnesses:
  pi_agent:
    # Run the installed Pi CLI.
    binary: pi
    # `--print` makes Pi run once and exit.
    prefix_args: ["--print"]
    # Pass the Dagu prompt as the final positional argument.
    prompt_mode: arg
    prompt_position: after_flags
    # Rename Dagu `with` keys to exact Pi flags where needed.
    option_flags:
      # `provider` is reserved by Dagu, so use `ai_provider` for Pi's LLM provider.
      ai_provider: --provider
      no_context_files: --no-context-files
      no_session: --no-session
      no_tools: --no-tools

steps:
  - id: summarize
    action: harness.run
    with:
      # Select the custom harness definition above.
      provider: pi_agent
      # Passed to Pi as `--provider openrouter`.
      ai_provider: openrouter
      # Passed to Pi as `--model openai/gpt-5.4-mini`.
      model: openai/gpt-5.4-mini
      # Small summarization task, so low reasoning is enough.
      thinking: low
      # Keep the run stateless and prevent file/tool access for this summary.
      no_session: true
      no_context_files: true
      no_tools: true
      prompt: |
        Summarize the README in one paragraph.

Generated invocation:

pi --print --provider openrouter --model openai/gpt-5.4-mini --no-context-files --no-session --no-tools --thinking low "Summarize the README in one paragraph."

Stdin

Pi print mode reads piped stdin and merges it into the initial prompt.

harnesses:
  pi_agent:
    # Run the installed Pi CLI.
    binary: pi
    # `--print` makes Pi read piped stdin and exit after one response.
    prefix_args: ["--print"]
    # Pass the Dagu prompt as the final positional argument.
    prompt_mode: arg
    prompt_position: after_flags
    # `provider` is reserved by Dagu, so expose Pi's flag as `ai_provider`.
    option_flags:
      ai_provider: --provider

steps:
  - id: summarize
    action: harness.run
    with:
      # Select the custom harness definition above.
      provider: pi_agent
      # Passed to Pi as `--provider openrouter`.
      ai_provider: openrouter
      # Passed to Pi as `--model openai/gpt-5.4-mini`.
      model: openai/gpt-5.4-mini
      # Small summarization task, so low reasoning is enough.
      thinking: low
      prompt: |
        Summarize these notes.
      stdin: |
        Import completed.
        Empty rows were skipped.

See Also

• Pi
• Custom Harness Definitions