# Type Analyze

> <!-- Generated from how_to_do.json by scripts/generate_skills.py. Edit how_to_do.json, not this file. -->

- Skill: `mihail-gribov/type-analyze` (Agent Skill)
- Install (CLI): `npx skillmds@latest add mihail-gribov/type-analyze`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mihail-gribov/type-analyze/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: mihail-gribov (https://skillmd.com/u/mihail-gribov)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mihail-gribov/type-analyze

---


<!-- Generated from how_to_do.json by scripts/generate_skills.py. Edit how_to_do.json, not this file. -->

# type-analyze

The user's request supplies the following values:

- `code_block` (required) - Python code for typing analysis

Placeholders such as `{code_block}` below refer to these values.

## Procedure

Analyze Python code in {code_block}:

1. Add type annotations:
   • function/method parameters
   • return values
2. Find untyped points: variables, parameters, results
3. For dictionaries with fixed keys — suggest TypedDict
4. For interface/structure abstraction — suggest Protocol (PEP 544)
5. If a value accepts multiple types — suggest Union
6. For containers: use Generic (`List[str]`, `Dict[int, Any]`)

Output format:
1. Code with added annotations (minimal, without logic)  
2. List of issues: missing annotations  
3. Recommendations: configure mypy/pyright for strict checking and Ignore‑flags

Logic in code doesn't change — only typing and advice.