David Lee
Community@davidsunglee
David Lee maintains a structured software delivery skill suite covering spec writing, planning, TDD, code review, debugging, and release management for git-based development.
Agent Skills by David Lee
Showing 17 vetted skills indexed across 1 GitHub repositories.
davidsungleerelease
Orchestrates versioned package releases with security audits, release notes, and user-run publish steps.
davidsungleecommit
Creates git commits with Conventional Commits-style subjects from staged changes.
davidsungleeusing-git-worktrees
Creates isolated git worktrees with directory selection, ignore verification, and baseline test validation.
davidsungleerefine-plan
Automates iterative plan review and edit cycles via a plan-refiner subagent until approval.
davidsungleetest-driven-development
Implements features and bug fixes using behavior-focused red-green-refactor test-driven development.
davidsungleerefine-code
Automates iterative code review and remediation loops via dispatched reviewer and coder subagents.
davidsungleesystematic-debugging
Diagnose bugs through four-phase root cause investigation before implementing fixes.
davidsungleeexecute-plan
Executes structured implementation plans via orchestrated coder, verifier, and test-runner subagents.
davidsungleegenerate-plan
Generates structured implementation plans from ideas, specs, or freeform descriptions via a planner subagent.
davidsungleescout
Dispatches a subagent to write structured codebase reconnaissance briefs to docs/briefs.
davidsungleefinishing-a-development-branch
Guides merging, PR creation, keeping, or discarding a completed development branch after verifying tests pass.
davidsungleefastlane
Orchestrates lightweight spec-driven implementation with a single coder subagent in the current workspace.
davidsungleerequesting-code-review
Dispatches a code-reviewer subagent with git diff context to assess production readiness.
davidsungleereceiving-code-review
Verifies code review feedback against codebase reality before implementing changes.
davidsungleeverification-before-completion
Enforces running verification commands and reading full output before claiming work is complete.
davidsungleedefine-spec
Writes structured specs to docs/specs/ via interactive Q&A with subagent dispatch and commit gating.
davidsungleeworkflow-skill-writing
Guides editing, simplifying, and refactoring pi-flow workflow skills without changing agent behavior.
Frequently Asked Questions About David Lee
FAQPage SchemaWhat tasks can I accomplish with David Lee's skill suite?▼
You can write structured specs, generate and execute implementation plans, run test-driven development cycles, perform systematic root-cause debugging, request and receive code reviews, manage git worktrees and branches, and drive full single-package or monorepo releases with security audits and readiness gates.
Who is the target audience for these skills?▼
Software engineers and development teams practicing disciplined delivery: spec-first planning, behavior-focused TDD, evidence-based verification, and gated code review before merging to main. It suits both solo developers and teams working in monorepos or multi-package repositories.
How does the release process work in practice?▼
The release skill drives versioning through a todo checklist, derives release notes from commits with manual review, runs a full security audit, enforces readiness gates, and leaves publish, push, tag, and GitHub release steps under explicit user control.
What prerequisites or repository structure do these skills expect?▼
Skills assume a git repository with conventions like docs/plans/, docs/specs/, and docs/briefs/ for artifacts. Some skills dispatch subagents and optionally use a terminal multiplexer pane, falling back to the main agent when unavailable.
How do the planning and review loops ensure quality?▼
generate-plan and refine-plan alternate review and edit cycles until approval or budget exhaustion, with commit gates on artifacts. refine-code and requesting-code-review dispatch reviewers with git diff context, while verification-before-completion requires running commands and reading full output before success claims.