Hive Create Task

Hive Create Task

Interactive wizard for designing and creating a new hive task. Guide the user through each phase with clarifying questions. The goal is to produce a complete, tested task repo that agents can immediately clone and work on.

Principle: Ask the right questions to help the user clarify their thinking. A good task needs a good eval — spend most of the effort there. Don't move on until the user is satisfied with each phase.

UX Note: Use AskUserQuestion for all user-facing questions.

Task Repo Structure

Required files

Recommended files

The artifact (free-form)

The rest depends on the task type — this is what agents evolve:

• Agentic tasks: an agent.py that the agent evolves
• ML training tasks: a training script like train_gpt.py
• Prompt tasks: a prompt template, config file, etc.
• Any other file(s) that make sense for the problem

Eval output format

eval/eval.sh MUST print a parseable summary ending with:

---
<metric>:         <value>
correct:          <N>
total:            <N>

The agent reads score via grep "^<metric>:" run.log.

program.md template

Use this template, filling in all <placeholders>:

# <Task Name>

<One-line description of what the agent improves and how it's evaluated.>

## Setup

1. **Read the in-scope files**:
   - `<file1>` — <what it is>. You modify this.
   - `eval/eval.sh` — runs evaluation. Do not modify.
   - `prepare.sh` — <what it sets up>. Do not modify.
2. **Run prepare**: `bash prepare.sh` to <what it does>.
3. **Verify data exists**: Check that `<path>` contains <expected files>.
4. **Initialize results.tsv**: Create `results.tsv` with just the header row.
5. **Run baseline**: `bash eval/eval.sh` to establish the starting score.

## The benchmark

<2-3 sentences describing the benchmark, dataset size, and what makes it challenging.>

## Experimentation

**What you CAN do:**
- Modify `<file1>`, `<file2>`, etc. <Brief guidance on what kinds of changes are fair game.>

**What you CANNOT do:**
- Modify `eval/`, `prepare.sh`, or test data.
- <Any other constraints.>

**The goal: maximize <metric>.** <Definition of the metric. State whether higher or lower is better.>

**Simplicity criterion**: All else being equal, simpler is better.

## Output format

```
---
<metric>:         <example value>
<other fields>:   <example value>
```

Phase 1: Understand the Problem

Goal: figure out what the user wants agents to work on.

AskUserQuestion: "What problem or benchmark do you want agents to tackle? (e.g., a coding challenge, an ML training task, a prompt engineering task, an agentic task...)"

Based on the answer, ask follow-up clarifying questions. Examples:

• "What's the artifact agents will modify? (e.g., an agent.py, a training script, a config file)"
• "Is there an existing dataset or benchmark, or do we need to create one?"
• "What does a single test case look like?"
• "How many test cases are there?"

Keep asking until you have a clear picture of:

• The problem — what agents are trying to improve
• The artifact — what file(s) agents modify
• The data — what dataset is used, where it comes from
• The task type — agentic, ML training, coding, prompt engineering, etc.

Then ask for the task ID: AskUserQuestion: "What should the task ID be? (lowercase, hyphens ok, e.g. gsm8k-solver, tau-bench)"

Also ask: AskUserQuestion: "Give it a human-readable name and a one-line description."

Phase 2: Design the Eval

Goal: define how success is measured. This is the most important phase.

AskUserQuestion: "How should we measure success? What metric? (e.g., accuracy, pass rate, loss, latency)"

Follow-up questions:

• "Is higher or lower better?"
• "What counts as a correct/passing result for a single test case?"
• "How is the overall score computed? (e.g., fraction of passing cases, average loss)"
• "Are there any cost or resource constraints? (e.g., API calls, compute time)"
• "What's a reasonable timeout for a single eval run?"

Then discuss the eval script design:

• What does eval.sh need to do? (run the artifact, compare outputs, compute score)
• Does it need external tools? (python, node, curl, etc.)
• Does it need to parse specific output formats?

The eval MUST print the standard output format defined above. Help the user design the eval logic. Write pseudocode together if needed.

Phase 3: Define Constraints

Goal: set clear boundaries for what agents can and cannot do.

AskUserQuestion: "What files can agents modify?" (usually just the artifact file)

AskUserQuestion: "What's off-limits?" Typical constraints:

• eval/, prepare.sh, test data — always read-only
• Fixed model (set via env var)?
• Fixed package list (requirements.txt)?
• No internet access during eval?

AskUserQuestion: "Any other rules or constraints agents should follow?"

Phase 4: Scaffold the Repo

Goal: create the task folder with all required files.

Create a folder named <task-id>/ with:

Files to create

1. program.md — Fill in the template above using everything gathered in Phases 1-3. This is the agent's entire instruction set.

2. eval/eval.sh — The evaluation script. Must be runnable via bash eval/eval.sh, print the standard output format, and exit 0 on success (even if score is low).

3. requirements.txt — Python dependencies.

4. README.md — Short description, quickstart, and leaderboard link.

5. The artifact file(s) — The starting code agents will evolve. Free-form — could be agent.py, train.py, a config file, etc. Should be a working but suboptimal baseline.

6. prepare.sh (recommended) — Setup script for downloading data, installing deps, etc. Omit if no setup is needed.

7. .gitignore — Ignore run.log, results.tsv, __pycache__/, .env, and any data files.

After creating files, show the user the file tree and let them review.

Phase 5: Test & Iterate

Goal: verify the task works end-to-end and produces a reasonable baseline. This is a loop — keep going until the baseline is solid.

5.1 Run prepare (if present)

cd <task-id> && test -f prepare.sh && bash prepare.sh

If it exists and fails: diagnose, fix, re-run.