text_input_playground.md

text_input_playground.json

Use this as a simple text-input example you can copy into an empty folder and hatch locally.

{
  "version": "2026-03-03.r1",
  "inputs": [
    {
      "type": "text",
      "text": "Summarize the meeting notes in one sentence."
    }
  ],
  "agent_schema": {
    "type": "object",
    "properties": {
      "summary": {
        "type": "string",
        "description": "One-sentence summary of the notes."
      }
    }
  },
  "actions": [
    {
      "name": "print_summary",
      "logic": {
        "!=": [
          { "var": "summary" },
          ""
        ]
      },
      "run": [
        {
          "kind": "exec",
          "program": "echo",
          "args": ["Summary created."]
        },
        {
          "kind": "exec",
          "program": "echo",
          "args": [{ "var": "summary" }]
        }
      ]
    }
  ]
}

Save it as text_input_playground.json, then run:

cargo ai hatch text_input_playground.json
./text_input_playground

For Windows users, run text_input_playground.exe or just text_input_playground.

You can also override the baked input at runtime:

./text_input_playground --input-text "This was a long meeting about launch timing, customer rollout, support staffing, and pricing changes. Summarize the most important takeaway in one sentence."

If you want the matching explanatory sidecar file next to the JSON, save this as text_input_playground.md:

# `text_input_playground`

## Purpose
- Summarize meeting notes in one sentence.

## Inputs
- One text input with the notes to summarize.

## Output
- `summary`: one-sentence summary of the notes.

## Action Flow
- If `summary` is not empty, run `echo "Summary created."`
- Then print the actual `summary` value.

## Local Loop
- Hatch with `cargo ai hatch text_input_playground.json`
- Run `./text_input_playground`
- Override the baked text later with `./text_input_playground --input-text "..."`