Stephen J Newhouse
Community@snewhouse · Earth
Data Scientist, Bioinformatician, Dad, Husband, MS warrior
Agent Skills by Stephen J Newhouse
Showing 19 vetted skills indexed across 1 GitHub repositories.
token-compression
Reduce AI agent output tokens with mode-based compression rules for AA-MA workflows.
complexity-router
Score task complexity and route high-risk work to architectural review.
understand-codebase
Generate an evidence-backed ONBOARDING.md guide for inherited codebases.
Dispatching Parallel Agents
Dispatch parallel AI agents to investigate and fix independent failures concurrently.
aa-ma-execution
Execute AA-MA plans from .claude/dev/active/ with validation and provenance logging.
prototype
Validate design questions with throwaway runnable prototypes.
debugging-strategies
Diagnoses bugs and performance issues using a reproduce-hypothesize-verify workflow across codebases.
plan-verification
Verify AA-MA plans by auditing claims, assumptions, and acceptance criteria.
goal-condition-synthesis
Synthesize falsifiable Claude Code /goal completion conditions from AA-MA plan artifacts.
agent-teams
Orchestrate multi-agent collaboration with lifecycle phases, quality gates, and structured state.
write-a-skill
Create Claude Code agent skills with SKILL.md and progressive disclosure.
verify-impl
Audit implementation diffs for completed milestones and gate critical findings.
grill-with-docs
Interview an LLM plan against project domain documentation and code.
impact-analysis
Analyze dependency impact before editing code to prevent breaking changes.
retro
Convert git commit history into weekly engineering retrospective metrics.
system-mapping
Map architecture, execution flow, logging, dependencies, and environment configuration before code changes.
defense-in-depth
Enforce layered validation across API boundaries, business logic, environment guards, and debug instrumentation.
aa-ma-plan-workflow
Run a five-phase planning workflow with validation gates for complex feature requests.
operational-constraints
Enforce operational constraints for LLM agent workflows at session start.