Agent Skills by nhouseholder
Showing 66 vetted skills indexed across 1 GitHub repositories.
content-research-writer
Automate research-backed writing with source gathering, outlines, and cited drafts.
git-sorcery
Generate commit messages from diffs and guide conflict resolution.
process-monitor
Monitor background processes for hangs, port conflicts, and zombies.
command-center
Decompose complex tasks into parallel workstreams and coordinate specialist subagents.
using-git-worktrees
Create isolated git worktrees for feature branches with setup checks.
qa-gate
Coordinate automated QA testing with tiered validation and pass/fail reporting.
senior-dev-mindset
Plan end-to-end feature implementations covering frontend, backend, deployment, and testing.
skill-manager
Cap active skills per message and resolve conflicts with priority rules.
fix-loop
Automate detection, diagnosis, and repair of failing tests in a CI loop.
smart-clarify
Resolve ambiguous user intent with 2-4 option multiple-choice questions.
test-driven-development
Enforce a test-first Red-Green-Refactor workflow for code changes.
iterative-retrieval
Refines codebase context for subagents via iterative 4-phase search loops with relevance scoring.
systematic-debugging
Guide engineers through a four-phase root-cause debugging workflow.
context-hydration
Load relevant project files and context before modifying code.
coding-standards
Enforce coding standards across TypeScript, JavaScript, React, and Node.js projects.
reflexion
Run opt-in reflect or critique modes for AI output self-review.
intent-detection
Detect user intent from plain language and route requests to the correct skill.
screenshot-dissector
Analyze screenshots to identify visual bugs with a pixel-level protocol.
deploy
Automate Cloudflare Pages and Workers deployment pipelines with linting, testing, and rollback.
dispatching-parallel-agents
Orchestrates parallel execution of independent tasks by isolated AI agents.
confusion-prevention
Stop work, capture state snapshots, and re-orient when confusion is detected.
think-efficiently
Enforce pre-action reasoning and action pruning to reduce token waste.
deep-research
Identify knowledge gaps and trigger authoritative research before coding.
model-router
Classify tasks and route them to Opus or Sonnet models by complexity.