Mimo Skill Creator

Skill Creator — MiMo Code

Create skills that extend MiMo Code with specialized workflows. Leverages task, bash, read, write, edit, skill tools natively.

Three Gates (Pre-flight)

All must pass before starting. If any is no → stop.

1. MiMo Code can't already do this well? → Skill is overhead.
2. User will use it 5+ times? → One-shot → direct prompt.
3. Model has it built-in? → Skill adds complexity, not value.

Skill Anatomy

skill-name/
├── SKILL.md          # Required — frontmatter + instructions
├── scripts/          # Optional — executable code
├── references/       # Optional — loaded on demand
└── assets/           # Optional — templates, icons, fonts

Frontmatter — only 3 fields are read by MiMo Code:

name: kebab-case       # Required. Letters, digits, hyphens. Max 64 chars.
description: >         # Required. Max 500 chars. Triggering conditions ONLY.
  Use when [trigger], [trigger], or [symptom].
hidden: true           # Optional. true = loaded but not in available_skills list.

Placement — MiMo Code discovers SKILL.md recursively from:

• Project: .mimocode/skills/**, .claude/skills/**, .agents/skills/**, .codex/skills/**
• Global: ~/.config/mimocode/skills/**, ~/.claude/skills/**, ~/.agents/skills/**
• Config: skills.paths and skills.urls in mimocode.json

The #1 Mistake: Description Trap

When description summarizes the workflow, the model follows the description and skips the body.

# ❌ BAD: Summarizes workflow → model takes shortcut
description: Use for TDD — write test first, watch it fail, write minimal code

# ✅ GOOD: Triggering conditions only → forces model to read the body
description: Use when implementing features or bugfixes, before writing code

Formula: [Action verb] + [value]. Use when [trigger 1], [trigger 2], ... Include 5+ triggers. Add exclusions. Be slightly "pushy".

Creation: 7 Steps

1. Classify Complexity

| Tier | SKILL.md | Dirs | |------|----------|------| | Simple | many mediocre

• ALWAYS/NEVER in caps → yellow flag → reframe with reasoning
• Context window is shared — every token must earn its place
• Cross-reference by name (compose:tdd), not @link (force-loads, burns context)
• Target: redundant content \n\n\n## Task\n"

})

Verify baseline differs. Do NOT tell subagent it's being tested.

### 7. Iterate

Improve → re-test → repeat until user satisfied or progress stalls.

## Evaluation Pipeline

### Architecture

MiMo Code subagents return results in response but do NOT reliably write to disk. **Controller persists all results.** Use `spawn` for parallel eval, `run` for blocking graders.

### Flow

**1. Test cases** — 10-20 prompts, 7:2:1 ratio (common/edge/anomalous):
```json
[{"id": 0, "prompt": "user task", "expectations": ["includes X"]}]

2. Paired subagents — batch 2-3 cases (4-6 concurrent):

task({ action: "spawn", agent: "general", description: "eval-0-with",
       prompt: "\n\nTask: " })
task({ action: "spawn", agent: "general", description: "eval-0-without",
       prompt: "Solve this task without any external instructions.\n\nTask: " })

3. Persist on notification — immediately write responses:

evals/iter-N/eval-/with_skill.md
evals/iter-N/eval-/without_skill.md

4. Grade — run blocking grader per case:

{"expectations": [{"text": "...", "with_skill": "pass", "without_skill": "fail", "evidence": "..."}]}

5. Aggregate via bash + Python → benchmark.md:

| Metric | with_skill | without_skill | Delta |
| Pass rate | 85% ± 5% | 35% ± 8% | +50% |

6. Analyze — non-discriminating (always pass), flaky (high variance), broken (always fail both).

Metrics

| Metric | Target | |--------|--------| | Routing accuracy | > 90% | | Output usability | > 80% | | Redundant tokens | && zip -r my-skill.skill my-skill/ \ -x "my-skill/evals/" -x "my-skill/iter-/" -x "__pycache__*"

## 5 Common Failures

| # | Failure | Fix |
|---|---------|-----|
| 1 | Description too vague | 5+ trigger phrases |
| 2 | SKILL.md dumping ground (800+ lines) | < 500 lines, split to references/ |
| 3 | ALWAYS/NEVER caps | Explain reasoning |
| 4 | No smoke test | Run Step 6 before shipping |
| 5 | Description summarizes workflow | Triggering conditions ONLY |

## Meta-Advice

1. **Description is everything.** Great skill + bad description = invisible.
2. **Explain why.** Model generalizes better with reasoning.
3. **Draft then look fresh.** First draft: too detailed or too vague.
4. **Generalize from feedback.** Encode the principle, not the fix.
5. **Context window is shared.** Every token earns its place.
6. **Test before deploying.** 15 min testing saves hours debugging.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [taxueseek](https://github.com/taxueseek)
- **Source:** [taxueseek/skill-optimizer](https://github.com/taxueseek/skill-optimizer)
- **License:** MIT

Install and usage instructions live in the source repository linked above.