Derek Thurn
Community@thurn · Mountain View, CA
Derek Thurn publishes 22 skills for git worktree orchestration, backlog execution, adversarial test validation, design documentation, and Playwright browser inspection.
Agent Skills by Derek Thurn
Showing 20 vetted skills indexed across 1 GitHub repositories.
adversarial-test-validation
Injects temporary user-visible bugs to verify test suites catch real regressions.
independent-review
Delegates a read-only code review to a fresh subagent and verifies each finding against the source.
wtbatch
Orchestrates parallel subagents in isolated git worktrees to execute batches of work items.
backlog
Convert rough bug lists into standalone task files through structured user interviews.
reddit-digest-to-epub
Convert AskHistorians Reddit digest URLs into self-contained EPUB ebooks via Playwright browser scraping.
wtm
Implements tasks directly on the master or main branch without creating worktrees.
wtcleanup
Replay worktree branch commits onto master as linear history, then remove the worktree and branch.
backlog-execute-serial
Executes backlog task files sequentially on one shared git branch using implementer and reviewer subagents.
backlog-execute
Executes backlog task files in parallel git worktrees with implementer-reviewer subagent pipelines.
batch-edit
Applies uniform transformations to many file entries using parallel subagents on split chunks.
enact
Orchestrates multi-agent execution of design documents through coordinated subagent pipelines.
wt-sequence
Executes multi-task implementation plans as stacked commits with overlapping review and speculative work.
technical-design-docs
Write standalone technical design documents, RFCs, and architecture proposals in Markdown.
scenario-design-docs
Generates technical design documents and QA scenario appendices through one-question-at-a-time interviews.
grill-me
Interviews the user question-by-question to stress-test plans and resolve design decisions.
implement-plan
Implements task-based plans through orchestrator and implementer Codex tasks using wt worktrees.
enact-audit
Verifies completed projects against design documents via blind review, QA, and remediation subagents.
wt
Implements tasks in isolated Tollgate git worktrees with speculative CI validation and gated promotion.
visual-review
Review running UIs through screenshots and rendered-state measurements to report bugs and design feedback.
Answers all questions in mixed prompts before requesting approval to implement.
Frequently Asked Questions About Derek Thurn
FAQPage SchemaWhat tasks can I accomplish with Derek Thurn's skills?▼
You can orchestrate multi-task implementation across isolated git worktrees, groom rough bug lists into standalone task files, run adversarial mutation testing against your test suite, obtain independent second-opinion code reviews, write RFC-style design documents, perform visual UI review, and inspect Playwright trace files from the command line.
Who are these skills designed for?▼
Software engineers and technical leads managing large backlogs, multi-step implementation plans, or design documentation. They suit developers who want parallel subagent execution with review gates, linear git history promotion, and structured validation before declaring work complete.
How do the worktree-based execution skills run in practice?▼
Skills like wt, wtbatch, and backlog-execute create Tollgate-managed git worktrees per task, run implementer plus reviewer subagent pipelines, then cherry-pick approved commits onto a linear integration branch. wtcleanup replays unique commits onto master and removes the worktree afterward.
What prerequisites or dependencies do these skills require?▼
Most skills require a local git repository with worktree support and subagent orchestration. The Playwright skills require npx, npm, and playwright-cli access. The reddit-digest-to-epub skill explicitly works without Reddit OAuth or API credentials.
Do these skills support design and planning before implementation?▼
Yes. technical-design-docs and scenario-design-docs produce standalone RFCs and architecture proposals with QA-ready scenario appendices, while grill-me interviews you relentlessly to stress-test plans, and qq ensures all questions are answered before any implementation begins.