Planning & Workflow

Run quality gates, stage specific files and write a clear commit message

Track project to-dos as GitHub issues with duplicate checks and closing notes

Explore requirements and design options through dialogue before building anything

Fix a GitHub issue end to end, from analysis through to a submitted pull request

Stress-test a plan or design with pointed questions that expose hidden assumptions

Pass design thinking findings into research and planning handoffs

Stage chosen paths and create one Conventional Commit or a ready message

Merge or rebase branches, resolve conflicts and push when you approve

Audit git identity, editor and diff tool settings and propose safe fixes

Manage GitLab merge requests and pipelines from a Python command line tool

Generate commit history and diff summaries between branches for pull requests

Prepare and open or update GitHub pull requests with preflight checks

Pressure-test a plan or decision with skeptical questions before acting

Follow an approved implementation plan, check off work and keep a changes log

Create or resume evidence-based implementation plans with optional critique

Assess an implementation plan for credibility without editing it

Gather task evidence and alternatives into dated research notes before planning

Draft design docs with diagrams, data models and trade-off options before coding

Walk through red-green-refactor with a failing test, minimal code and cleanup

Add go or no-go verification gates between project phases to catch errors early

Ask the deep questions behind a technical decision and write a build spec

Build a feature from an approved spec, one slice at a time, following project docs

Improve an existing app using its code, past work and competitor evidence

Find unpushed commits, dirty worktrees and stashes before cleaning up git state

Open upstream pull requests that follow CONTRIBUTING rules with minimal diffs

Operate GitHub repos, PRs, issues and Actions via gh, REST and GraphQL with readback

Find and verify existing code and past decisions before building something new

Compare libraries, frameworks and build-vs-buy options with a gated checklist

Turn the winning evo experiment into a minimal, mergeable pull request

Clarify first, write tests first and protect review load on every coding task

Run issue-first checks and then open a pull request ready for review

Split oversized changes into chained PRs that keep each review focused

Guide receipt-driven defect fixes with review gates, lineage and bounded rounds

Plan commits as reviewable work units that keep tests and docs with code

Create a GitHub App so the assistant can push commits and open pull requests

Measure engineering health with Core 4 and manage tech debt as leverage

Ship faster with ruthless scoping, rituals and technical guardrails

Map a codebase's structure, tech stack and conventions when onboarding

Pick architecture patterns and structure a project, from monoliths to microservices

Manage branches, commits, pull requests and merge conflicts in Git

Explore ambiguous features, architecture choices and designs before writing code

Execute a written implementation plan in stages with review checkpoints

Finish a development branch by merging, opening a PR or cleaning up its worktree

Review risky decisions from first principles to cut scope and needless complexity

Set a goal, success evidence and stop condition before starting a task

Grant bounded execution authority for a decision with freshness and scope checks

Compare candidate approaches on explicit criteria and show trade-offs, not one score

Record a bounded engineering decision as a decision record with its rationale

Generate three to five distinct candidate approaches with their weakest links visible