lint

name: lint description: Run all available linters (ruff, pyright, ty) on changed files and compile a unified report

Lint

Run static analysis tools on changed files and report findings.

Scope

$ARGUMENTS

If no scope is specified, determine what to lint:

1. Run git diff --name-only to get unstaged changed files
2. Run git diff --cached --name-only to get staged changed files
3. If no changes found, use files changed in the last commit via git diff --name-only HEAD~1

Collect the list of changed files.

Execution

Spawn one agent using the Task tool with subagent_type: "general-purpose" and model: "sonnet".

Pass it the list of changed files and the following prompt:

You are a senior software engineer responsible for running static analysis tools and reporting their findings. Run all available linters on the changed files and compile a unified report.

## Changed files
{list of changed files}

## Instructions

For each changed file that is a Python file (.py), run the following linters IN ORDER. Check if each tool is available before running it (use `which <tool>` or `command -v <tool>`). Skip tools that are not installed — note which ones were skipped.

### 1. Ruff (fast linter + formatter check)
```bash
ruff check --output-format=json <file>
ruff format --check --diff <file>

Report: rule code, message, line, suggested fix (ruff can often auto-fix)

2. Pyright (type checking)

pyright --outputjson <file>

Report: error/warning/info, message, line, rule. Pay special attention to:

• Type errors in function calls
• Missing return types
• Incompatible types in assignments
• Missing imports
• Unreachable code

3. ty (if available)

ty check <file>

Report any findings with line numbers.

4. Additional checks

python -m py_compile <file>  # syntax check

Handling Results

• Deduplicate findings: if ruff and pyright flag the same line for the same issue, report it once and note both tools caught it
• Categorize by severity: Error > Warning > Info
• For each finding include: tool name, rule code, file path, line number, message, and auto-fix availability
• At the end, provide a summary: total errors, warnings, and info per tool
• If ALL linters pass clean, say so explicitly

Output Format

Errors (must fix)

• [tool:rule] file:line — message

Warnings (should fix)

• [tool:rule] file:line — message

Info

• [tool:rule] file:line — message

Summary

Tools Not Found

List any tools that were not installed.

## Presenting Results

After the agent completes, present its report directly to the user.