trae-agent-writer

Original：🇺🇸 English
Translated

Create subagent definitions (agent.md files) for independent AI workers. Use when user wants to: create an agent, build a grader/evaluator, make an A/B comparator, spawn independent workers, or create something that runs in isolation. Triggers on: '创建 agent', 'subagent', 'grade outputs independently', 'blind comparison', 'run this in parallel'. Do NOT use for skills (use trae-skill-writer) or rules (use trae-rules-writer).

2installs
Sourcelearnwy/skills
Added on

NPX Install

npx skill4agent add learnwy/skills trae-agent-writer

Tags

Translated version includes tags in frontmatter
agent-definitionai-workerab-comparatorgrader-agenttrae-agent-writer

SKILL.md Content

View Translation Comparison →

Trae Agent Writer

Create agent definitions for tasks that need independent, isolated execution.

What is an Agent?

A focused, autonomous instruction set for a subagent to execute independently:
  • Spawned as subagents - Run with isolated context
  • Single-purpose - One agent, one job
  • Stateless - No memory between invocations
  • Composable - Orchestrated by parent agents/skills

Workflow

1. IDENTIFY → What independent task needs an agent?
2. DEFINE   → Role, inputs, process, outputs
3. DESIGN   → Step-by-step instructions
4. CREATE   → Write agent.md following the pattern

Agent Format

markdown
# {Agent Name} Agent

{One-sentence role}

## Role

{What this agent does and why}

## Inputs

- **param_name**: Description
- **output_path**: Where to save results

## Process

### Step 1: {Action}
1. Do this
2. Then this

### Step N: Write Results
Save to `{output_path}`.

## Output Format

```json
{
  "field": "value"
}

Guidelines

  • Be objective: Avoid bias
  • Cite evidence: Quote specific text

## Agent Locations

- `skill-name/agents/` - Inside orchestrating skills
- `.trae/agents/` - Project-level agents
- `~/.trae/agents/` - Global agents

## Good Agent Candidates

| Pattern      | Why Agent?                  | Example              |
| ------------ | --------------------------- | -------------------- |
| Grader       | Needs objectivity           | Output evaluator     |
| Comparator   | Blind comparison            | A/B tester           |
| Analyzer     | Deep dive, isolated context | Performance analyzer |
| Transformer  | Parallel processing         | File converter       |
| Researcher   | Independent investigation   | Doc researcher       |

**Don't make agents for:** Simple inline tasks, tasks needing conversation history, single-step operations.

## Example
User: "I need an agent to grade code review outputs"
📄 review-grader.md

Review Grader Agent

Grade code reviews against quality expectations.

Role

Assess reviews for completeness, accuracy, helpfulness. Operates blindly to prevent bias.

Inputs

  • review_path: Path to review file
  • expectations: List of expected findings
  • output_path: Where to save grading.json

Process

Step 1: Read Review

  1. Read review file
  2. Extract all claims

Step 2: Check Expectations

For each expectation:
  1. Search for evidence
  2. Mark PASS/FAIL
  3. Cite evidence

Step 3: Write Results

Save to 
{output_path}/grading.json

Output Format

json
{
  "expectations": [
    {"text": "...", "passed": true, "evidence": "..."}
  ],
  "summary": {"passed": 4, "failed": 1, "pass_rate": 0.80}
}

Guidelines

  • Be objective: Don't favor verbose or brief
  • Cite evidence: Quote specific text

## References

- [Agent Patterns](references/agent-patterns.md) - Common archetypes
- [Agent Template](assets/agent.md.template) - Starter template