Pre-PR Check Skill

Run comprehensive local CI checks before PR creation. Auto-detects project tooling, runs all checks, and offers interactive fixes on failure.

Quick Start

# Run all detected checks
/sc:pr-check

# Run specific check categories
/sc:pr-check --only lint,test

# Quick mode - skip slow checks (e2e, security)
/sc:pr-check --quick

# Fix mode - auto-fix what's possible
/sc:pr-check --fix

# Dry run - show what would be checked
/sc:pr-check --dry-run

Behavioral Flow

Detect - Identify project tooling (language, linter, formatter, test runner, security scanner)
Plan - Build check sequence based on detected tools
Execute - Run checks sequentially, streaming output
Remediate - On failure, offer interactive fixes per check type
Re-run - After fixes, re-run all checks until green or user stops
Report - Summary of results and fixes applied

Flags

Flag	Type	Default	Description
`--only`	string	all	Comma-separated: `lint`, `format`, `test`, `security`, `types`, `e2e`, `build`
`--quick`	bool	false	Skip slow checks (e2e, security scan)
`--fix`	bool	false	Auto-fix fixable issues without prompting
`--dry-run`	bool	false	Show detected checks without running
`--strict`	bool	false	Treat warnings as errors

Phase 1: Project Tooling Detection

Auto-detect the project's tooling by examining config files and package manifests.

Detection Matrix

Check Type	Python	JavaScript/TypeScript	Go	Rust
Linter	ruff, flake8, pylint	eslint, biome	golangci-lint	clippy
Formatter	ruff format, black	prettier, biome	gofmt	rustfmt
Types	mypy, pyright	tsc	(built-in)	(built-in)
Tests	pytest, unittest	jest, vitest, mocha	go test	cargo test
Security	bandit, pip-audit	npm audit, snyk	govulncheck	cargo audit
Build	-	tsc, vite build	go build	cargo build
E2E	-	playwright, cypress	-	-

Detection Strategy

# Python detection
[ -f "pyproject.toml" ] || [ -f "setup.py" ] || [ -f "requirements.txt" ]

# JavaScript/TypeScript detection
[ -f "package.json" ] || [ -f "tsconfig.json" ]

# Go detection
[ -f "go.mod" ]

# Rust detection
[ -f "Cargo.toml" ]

# Check for specific tools in config files
grep -q "ruff" pyproject.toml 2>/dev/null      # Ruff
[ -f ".eslintrc*" ] || grep -q "eslint" package.json 2>/dev/null  # ESLint

Detection Output

## Project Tooling Detected

| Category | Tool | Config | Status |
|----------|------|--------|--------|
| Language | Python 3.12 | pyproject.toml | OK |
| Linter | Ruff | pyproject.toml [tool.ruff] | OK |
| Formatter | Ruff Format | pyproject.toml [tool.ruff] | OK |
| Type Check | mypy | pyproject.toml [tool.mypy] | OK |
| Tests | pytest | pyproject.toml [tool.pytest] | OK |
| Security | bandit | .bandit | OK |
| E2E | playwright | playwright.config.ts | OK |

Running 7 checks...

Phase 2: Check Execution

Run checks in dependency order:

Step	Check	Rationale
1	Linter	Catch syntax/style errors first (fastest)
2	Formatter	Ensure consistent formatting
3	Type checker	Catch type errors before runtime
4	Unit/Integration tests	Verify correctness
5	Security scan	Catch vulnerabilities
6	Build	Verify compilation
7	E2E tests	Full integration (slowest, skipped with --quick)

Execution Pattern

For each check:

Display step number and check name: [1/7] Running Ruff linter...
Execute the command with timeout
Capture exit code and output
On success: show green checkmark, move to next
On failure: enter remediation flow

Phase 3: Interactive Remediation

When a check fails, identify the failure type and offer targeted fixes.

Lint Failures

[1/7] Ruff linter... FAILED (3 errors)

  src/api/routes.py:15:1 F401 unused import
  src/models/user.py:8:5 E712 comparison to True
  src/utils/helpers.py:22:80 E501 line too long

Actions:
  [f] Auto-fix with `ruff check . --fix` (fixes 2 of 3)
  [m] Manually fix remaining issues
  [s] Skip and continue
  [q] Quit

Test Failures

[4/7] Tests... FAILED (2 failures)

  FAILED tests/test_api.py::test_create_user - AssertionError
  FAILED tests/test_auth.py::test_login_expired - TimeoutError

Actions:
  [a] Analyze and fix failing tests
  [r] Re-run failed tests only
  [s] Skip and continue
  [q] Quit

Security Failures

[5/7] Security scan... FOUND 2 issues

  HIGH: SQL injection risk in src/api/users.py:45 (B608)
  LOW:  Assert used in production code src/utils/debug.py:12 (B101)

Actions:
  [a] Analyze and remediate issues
  [i] Show details for each finding
  [s] Skip and continue
  [q] Quit

Fix Type Reference

Check	Auto-fixable	Manual Fix Needed
Lint	Most style issues	Logic errors, unused code decisions
Format	All formatting	None
Types	None	Type annotations, casts
Tests	Some assertion fixes	Logic bugs, mock updates
Security	Some patterns	Architecture changes
Build	Dependency installs	Code errors
E2E	Selector updates	Flow changes

Phase 4: Re-run After Fixes

After applying any fix:

Re-run all checks (not just the failed one)
Continue the fix loop until all pass or user quits
Track all fixes applied during the session

Phase 5: Results Report

## PR Check Results

**Status**: ALL PASSED | X FAILED
**Duration**: Ns

### Check Results

| Step | Check | Status | Duration |
|------|-------|--------|----------|
| 1/7 | Ruff linter | PASSED | 1.2s |
| 2/7 | Ruff formatter | PASSED | 0.8s |
| 3/7 | Type check | PASSED | 3.4s |
| 4/7 | Tests (48 passed) | PASSED | 12.1s |
| 5/7 | Security scan | PASSED | 2.3s |
| 6/7 | Build | PASSED | 5.6s |
| 7/7 | E2E tests | SKIPPED | --quick |

### Fixes Applied (if any)
- Auto-fixed 3 lint errors (ruff --fix)
- Fixed test assertion in test_create_user
- Remediated B608 SQL injection in users.py

### Next Steps
- Run `/sc:git commit` to commit changes
- Run `/sc:readme` to update documentation
- Create PR with `gh pr create`

MCP Integration

PAL MCP

Tool	When to Use	Purpose
`mcp__pal__debug`	Test failures	Root cause analysis for failing tests
`mcp__pal__codereview`	Security findings	Validate security remediation
`mcp__pal__precommit`	Final validation	Multi-model pre-commit check

PAL Usage Patterns

# Pre-commit validation after all checks pass
mcp__pal__precommit(
    path="/path/to/repo",
    step="Validating all changes before PR",
    findings="Lint, test, security, type check results",
    confidence="high"
)

# Debug complex test failure
mcp__pal__debug(
    step="Investigating intermittent test failure in test_auth",
    hypothesis="Race condition in async test setup",
    confidence="medium"
)

Rube MCP

Tool	When to Use	Purpose
`mcp__rube__RUBE_MULTI_EXECUTE_TOOL`	Notifications	Post results to Slack, update Jira

Tool Coordination

Bash - Execute linters, formatters, test runners, security scanners
Glob - Detect config files for tooling identification
Grep - Parse failure output, find config patterns
Read - Read failing test files, security findings
Edit - Apply auto-fixes

Related Skills

/sc:test - Detailed test execution and coverage
/sc:log-fix - Debug from log analysis
/sc:git - Git workflow operations
/sc:analyze - Deeper code quality analysis

sc-pr-check