silvio-l
UserMy personal Claude Code skills. Engineering process, context hygiene, and AI-agent maintenance — straight from my .claude directory.
Categories
Indexed Skills (8)
ratchet-up
Context-safe agentic loop that ratchets a feature's issues forward — one click at a time, never backward. Discovers ready issues in a local feature directory, respects blocked-by dependencies, optionally consults a planner for ordering, dispatches isolated workers and read-only reviewers, runs format/analyze before review, persists rework feedback on disk, commits approved issues with conventional commits, and stays on the active branch (no worktrees, no auxiliary branches). Project-agnostic — gate commands auto-detected from CLAUDE.md or .ratchet-up/config.yaml. Usage: /ratchet-up <feature-path> [max-iter]
seo-audit
Local-first, free-tier-only SEO audit for any repo with a built website. Phase pipeline: inventory (framework, pages, SEO assets, app-store listings, domain doc) → brand-consistency scan (anti-vocabulary table from CONTEXT.md/CLAUDE.md matched against built HTML, with contrastive-marker and frontmatter-flag suppression) → external probes (Lighthouse, pa11y, GSC, W3C, Schema, Observatory, PageSpeed) when `--url` is passed → synthesis (dedup, score by severity × user_impact / fix_effort, deterministic tiebreaker) → write Markdown report under .scratch/<feature>/seo-audit-<date>.md. Optional opt-in push module (`--push`): IndexNow + Bing Webmaster URL submission + llms.txt generator, confirmable per operation, with `--dry-run`. Use when the user says "SEO audit", "check the brand voice on the site", "scan dist for anti-vocabulary", "is our site consistent with CONTEXT.md", "prep an SEO report", "push URLs to IndexNow / Bing", "generate llms.txt", or runs /seo-audit.
to-roadmap
Erzeugt, aktualisiert und pflegt den Status einer agentenoptimierten Sprint-Roadmap in `.scratch/roadmap.md`. Drei Modi: `create` zerlegt ein Ideen-/Konzeptdokument (rohes PRD, Brainstorm) in ~100k-Token-Sprints mit stabilen Sprint-IDs; `update` modifiziert eine bestehende Roadmap nach freier Anweisung mit Diff-Plan vor dem Schreiben; `status` setzt den Lebenszyklus eines Sprints (todo → in-progress → done), aufrufbar manuell oder durch `ratchet-up`. Grobe Vorstufe vor `/to-prd` — ein Sprint = später ein PRD = später ein Bündel Issues. Use when user wants to create, edit, refine, restructure or status-mark an implementation roadmap or sprint plan. Trigger phrases — roadmap erzeugen, Roadmap aktualisieren, Sprint hinzufügen, Sprint entfernen, Sprint umschreiben, Sprint splitten, Roadmap-Status setzen, sprint status done, idea to roadmap, PRD zerlegen, Featureplanung.
apple-notes
Read, search, write, and extract images from Apple Notes on macOS via a single dispatcher script (AppleScript under the hood). Treats subfolders of one configured "company folder" as projects, auto-maps the current git repo to its project subfolder, and enforces a four-folder issue layout (inbox/ready/done/docs) with title-prefix conventions (BUG:/FEAT:/IDEA:/FB:/TECH:) so non-technical collaborators can drop in bug reports that agents can later triage and act on. Use when the user references Apple Notes, says "check my notes", "pull bug reports from Notes", "what did my partner write about this app", wants to extract screenshots from a note, triage the inbox, or move an issue between statuses.
context-optimization-audit
Audit and optimize Claude Code context bloat across skills, agents, commands, MCP/plugin config, project instructions, and duplicated context. Use when asked to reduce token usage, remove overlap, clean up skills, optimize Claude Code setup, or review loaded context.
domain-glossary
Collaboratively build and maintain the project's domain vocabulary, glossary, and key conceptual decisions stored in CONTEXT.md. Use when the user wants to create, review, or refine project terminology, domain language, glossary entries, or conceptual decisions — including updating CONTEXT.md or discussing what belongs in it.
full-quality-scan
Runs all configured linters and security scanners (cppcheck, ESLint, Semgrep, osv-scanner, dart analyze) across the entire repository, then fixes every finding — directly if ≤10, via a structured plan with parallel subagents if more. Use when user wants a full repo-wide quality or security scan, says "scan everything", "fix all lint issues", "clean up the whole repo", "full quality check", or asks to run tools across the full codebase (not just staged files or recent changes).
humanize-text
Scan text, markdown, Astro, HTML, and TypeScript files for AI-generated slop patterns in German and English. Three modes: scan (report findings, always exits 0), score (quality gate, exits 1 when below threshold), rewrite (script-assisted LLM pass to remove slop, only changes files after explicit user OK). The pipeline is deterministic — lexicon matches + scoring happen offline before any LLM involvement. Use when the user says "mach diesen Text menschlicher", "prüf den Absatz auf KI-Floskeln", "scan for AI tells", "check for slop", "humanize this text", "detect overused AI phrases", "run a slop score on this file", or runs /humanize-text.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.