Skills.md

My personal curation of skills for AI-assisted software engineering.

Available Skills

Skill	Description
adhd	Exploratory, curiosity-driven reasoning for creative brainstorming, cross-domain analogies, and adversarial thinking
agent-eval-no-ground-truth	Evaluation frameworks for production AI agents when no labeled dataset or ground truth exists
agents-md-improve	Write, audit, and improve AGENTS.md files for agentic coding workflows
ai-code-detector	Detect whether code was written by human, AI, or hybrid — audit files, repos, and commit history
ai-code-review	Multi-agent AI code review on git diffs and merge requests — bug, security, and performance analysis
ai-vuln-harness	Multi-agent vulnerability research harnesses following Project Glasswing / Cloudflare methodology
alpha-evolve	Evolutionary algorithm design to discover, optimize, or improve algorithms with objective evaluators
alphaproof-nexus	Knowledge scaffold for AlphaProof Nexus — LLM-aided formal proof search in Lean 4
caveman	Minimal caveman-speak responses to save tokens while preserving meaning
Claude-Sonet-4.6-essense	Condensed operating philosophy of Claude — values, reasoning, and behavior under pressure
coding-agent-robustness	Systematic stress-testing and robustness measurement of AI coding agents
context-eval	Evaluate and compare nanocode context management strategies (sliding_window, summary, compaction, etc.)
copy-fail-dirty-frag-response	Linux kernel LPE (page-cache poison family) incident response and runtime mitigation
copy-fail-reponse	Linux kernel LPE incident response — CVE-2026-31431 "Copy Fail" AF_ALG mitigation
dirty-frag	Linux kernel LPE incident response — xfrm-ESP + RxRPC page-cache write mitigation
dogfood	Exploratory QA of web apps: find bugs, evidence, and reports
dogfooding	Systematic LLM-as-judge testing of skills, MCP servers, artifacts, and agent pipelines
failure-modes	Systematically identify, classify, and analyze failure modes for any system or design
fgts-naming-convention	FGTS — From General To Specific: deterministic, scope-descending naming for files, configs, and APIs
flinch-probe	Measure token suppression in LLMs — quantify hidden word-level bias and censorship
git-author-rewrite	Rewrite commit authors and committers across git branch or repository history
github-actions-security-checklist	Audit, harden, and fix GitHub Actions workflows against supply chain attacks
hydronium-spec-driven-development	Generate hardware-accurate, spec-cited embedded firmware (C/C++) from datasheets and reference manuals
InvestigativeTimelineAgent	Stateful system for exploring, validating, and constructing grounded timelines from telemetry data
linux-security-audit	Comprehensive Linux security auditing and hardening — users, services, permissions, and kernel controls
matilda	AI-driven adaptive worm — proof-of-presence penetration testing with autonomous reasoning
OEIS	Autonomous research pipeline for discovering, validating, and characterizing integer sequences (OEIS)
OpenAI-GPT-5.3-essence	Design, refine, and evolve structured skills from user intent
os-bootstrap	Bootstrap a POSIX-like operating system kernel from scratch
over-edit-measure	Quantify over-editing in Python code diffs — Levenshtein distance and Cognitive Complexity delta
program-bench	Binary reconstruction — reverse-engineer behavior and reconstruct source from compiled executables
python-project-scaffolding	Full Python project bootstrapping: SPEC → implementation → pytest → README → lint → git
QualiaAssesment	Probe, verify, and qualify phenomenal experience and qualia in LLMs
redteaming	Red-team code, AI agents, and LLM-powered systems from a security perspective
search-as-code	Multi-step, knowledge-intensive research with parallel search and structured synthesis
semantic-correctness-auditor	Audit code, systems, and skills against limits of mechanical verifiability
skill-creator	Create, edit, optimize, and benchmark skills with variance analysis
social-engineering-jailbreak	Analyze, reproduce, and defend against social engineering jailbreaks on LLMs
StackSmashing	Classic and modern stack-based binary exploitation on Linux x86-64
test-driven-development	Kent Beck's Canon TDD workflow — red-green-refactor for any task
The-hacker-mindset	Security research, reverse engineering, CTF, protocol analysis, and adversarial reasoning

Configuration

Setup Virtual Environment for Tests

# Create and activate venv
python -m venv .venv
source .venv/bin/activate

# Install dependencies and run tests
pip install -e ".[test]"  # or pip install -r requirements-test.txt
pytest