spbt-skill：将git pull转化为结构化验证工作流的Claude Code智能技能

spbt-skill is a portable Claude Code and Cursor intelligent skill that converts simple git pull operations into a complete validation workflow. It addresses code merge validation pain points by automating steps like code stashing, commit classification, test generation, and AI验收验证, helping teams standardize post-merge verification processes.

In software teams, post-merge validation is critical but often overlooked. When developers run git pull, they face challenges: handling uncommitted local changes, identifying which commits need testing, ensuring new code doesn’t break existing features, and generating structured test docs. Traditional manual approaches are error-prone and hard to standardize—this is the problem spbt-skill solves.

spbt-skill’s workflow includes:

1. Safe Stash: Securely stash uncommitted changes with recorded references for recovery.
2. Pull Updates: Execute git pull (prioritize fast-forward; stop on merge conflicts).
3. Commit Classification: AI-powered classification into "application code changes" or "tool config changes".
4. Interactive Selection: Let developers choose which commits to test (tool config commits default to audit records).
5. Test Generation: Create manual checklists, security notes, Playwright/API scripts, Colab notebook units.
6. Env Var Management: Strict naming convention (UPPER_SNAKE_CASE_SCRUM_) for per-SCRUM工单 configs.

spbt-skill uses Agent Skill architecture (reusable workflow modules). It supports Claude Code and Cursor:

• Claude Code: Copy SKILL.md to .claude/skills/stash-pull-build-tests.
• Cursor: Copy SKILL.md to .cursor/skills/stash-pull-build-tests. Installation options:
• Project-level (recommended): ./scripts/install.sh --target <repo>.
• User-level: ./scripts/install.sh --user.
• Windows: Use PowerShell scripts (install.ps1 with -Target/-UserScope).

spbt-skill’s highlights:

1. Smart Commit Classification: Differentiates application code vs tool config changes to focus testing.
2. Mixed Validation Modes: Combines manual checklists, automated scripts, and Colab notebooks for flexibility.
3. AI-Assisted Validation: Uses AI for natural language triggering, targeted validation questions, and security risk identification.

spbt-skill benefits teams in multiple scenarios:

• Daily Dev Sync: Automates stash, pull, commit analysis, and test generation for daily updates.
• Code Review: Assists reviewers with commit classification, checklists, and security notes.
• Regression Testing: Generates Playwright/API scripts and Colab notebooks for frequent validation.
• Team Collaboration: Standardizes workflows, reduces new member learning curve, and ensures consistent test docs.

Current Limitations:

• Dependent on Claude Code/Cursor IDEs for full functionality.
• Environment variable conventions require team learning.
• Default templates may need project-specific adjustments.

Best Practices:

• Start with small pilot projects (1-2 SCRUM工单).
• Establish team agreements on test requirements for different commit types.
• Regularly update validation templates based on usage.
• Ensure sensitive env vars are not committed (use .gitignore for .env.stash-pull).

spbt-skill represents a new direction in dev workflow automation—combining human judgment with AI to automate validation planning and execution. It frees developers from tedious prep work, letting them focus on creative tasks. As AI evolves, more such skills will emerge to simplify dev processes and boost software quality. For teams aiming to improve code quality and standardize validation, spbt-skill is a thoughtful solution.