Design automated tests around real behavior, useful confidence, and maintainable contracts.
santos-sanz/ai-dev-workflow
SkillsMP has collected 16 skills from santos-sanz/ai-dev-workflow. Open a skill to review its source and details.
- Latest recorded source activity
- SkillsMP catalog refreshed
- skills collected
- 16
- GitHub stars
- 0
- GitHub forks
- 0
Skills in this repository
Showing 16 of 16 collected skills.
Design automated tests that reduce uncertainty, cover real contracts, and avoid brittle coverage theater.
Clarify underspecified, high-risk, or product-heavy work until scope, constraints, success criteria, and trade-offs are clear enough to plan or implement.
Diagnose bugs, flakes, regressions, and unexplained failures with reproduction, hypotheses, instrumentation, fix, and regression evidence.
Create a portable continuation summary when work must cross sessions, agents, branches, or tools and native memory is not enough.
Produce a decision-complete implementation plan before substantial edits, using native plan/read-only features where available.
Review a branch, PR, or working tree against the intended spec and repo standards, using native diff and review tools first.
Prepare completed work for commit, push, and pull request with scope review, final verification, PR story, assumptions, and risks.
Build features or fixes one behavior-first vertical slice at a time using the active agent's native edit, run, and verification tools.
Interview the user until goal, scope, constraints, success criteria, and trade-offs are clear enough to plan or implement.
Diagnose bugs, flakes, and regressions through reproduce, minimize, hypothesize, instrument, fix, and regression-test.
Produce a decision-complete implementation plan before code changes, including interfaces, data flow, verification, assumptions, and rollout notes.
Review a branch, PR, or working tree against the intended spec and the repo's documented standards.
Prepare completed work for a pull request with final verification, commit hygiene, push, and PR description.
Build features or fixes one behavior-first vertical slice at a time using a red-green-refactor loop.
Summarize active work so another agent or future session can continue without replaying the full conversation.