DavoDC
UserBudget-aware, self-improving Claude Code - built from hundreds of sessions of daily use
Categories
Indexed Skills (23)
aristotle
Aristotle First Principles Deconstructor - strip assumptions, find irreducible truths, rebuild from zero
cross-synth
Cross-synthesise any set of things (repos, files, docs, patterns, configs, tests, features) to find similarities, differences, gaps, and opportunities. General-purpose pattern-matching across multiple subjects. Like /deep-dive but for comparing N things sideways rather than one thing deeply.
deep-dive
Deep dive a topic, file, directory, or repo - think hard, investigate thoroughly, miss nothing
premortem
Pre-mortem risk analysis - imagine the plan already failed, work backward to find every reason why. Classifies risks as Tigers / Paper Tigers / Elephants, synthesises most likely failure, most dangerous failure, and hidden assumption, then produces a concrete revised plan. Output goes to terminal (like /aristotle), not TEMP.
prioritise
Rank any list or named backlog by leverage using Aristotle's "who benefits and are they a bottleneck?" lens. Returns a ranked list with one-line justification per item.
undo-commits
Undo the last N commits via git reset --soft, show staged diff, recommit cleanly. Safe - never rebases, never force-pushes.
checkpoint
Create, list, or verify named checkpoints during long sessions or loops
commit-all
Commit all files in git status to this repo - stages everything, writes commit message, no push
socrates
Socratic Rule Evaluator - examine every principle against 5 questions, delete zombies, strengthen weak ones, simplify verbose ones
survey-repo
Quick codebase summary - language, purpose, key files, tests, entry points, open TODOs
think
Full first-principles build workflow - Aristotle deconstruction then 5-step engineering algorithm then instantiation check. The three Core Thinking Principles in one invocable skill.
commit-chunks
Commit changed files in logical chunks - one commit per feature/fix/topic
step-commits
Plan changes as atomic commits upfront, then implement and commit one at a time
repo-status
Multi-repo status overview - branch, dirty file count, unpushed commits per repo. Read-only, never fetches.
human-voice
Audit and rewrite text to remove AI writing patterns - for letters, emails, messages, READMEs, or anything written to a person
make-public
Checklist to safely make a GitHub repo public - scan for secrets, check gitignore, verify README
process-feedback
Process a user feedback file with mandatory dual output - product tasks AND Claude learnings
refine-ideas
Clarify IDEAS.md priorities by asking ONE question per item - Claude-driven, derives tiers from answers, works with any format
reflection
Read recent session history, find patterns, update workspace config and memory. Delta-first - only reads sessions since last reflection.
release
Create a tagged GitHub release - detect version, write changelog, create git tag, draft release notes
skill-suggest
Context-aware skill suggestions - get relevant skills based on your task
validate-rules
Pre-commit validation - verify workspace rules are followed before pushing. Checks repo index, pending-actions discipline, scope documentation, TDD ratio, and rule file changes.
check-compact
Check context level using actual ctx% from statusline data
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.