#25Zavelinski/model-router1 skillsmodel-routerCost-aware model routing for multi-step work. Use when planning a task with mixed difficulty, when token/cost is a concern, or when orchestrating subagents. Decides which subtasks go to a cheap model (Sonnet/Haiku) and which need Opus, and sets the model per…2026年6月29日
#26Zavelinski/context-warden1 skillscontext-wardenRuntime defense against context rot. Use when a session grows long, before auto-compaction, when the agent starts re-reading files or losing track, or when token cost per turn climbs. Keeps the working set small via observation masking and compacts the…2026年6月29日
#27Zavelinski/claude-code-sync1 skillssyncWhen the user types "sync", check the locally installed Claude Code skills against the latest published version on GitHub and report drift; with the user's OK, sync outdated public skills. Always state the current tool version before claiming anything is up…2026年6月29日
#28Zavelinski/claude-code-safety-net1 skillssafety-netUse when the user wants to undo or roll back changes Claude made to files, recover a previous state, or turn on an automatic edit checkpoint net. Triggers - "undo", "desfazer", "voltar atras", "reverter as mudancas do Claude", "rollback", "restaurar", "safety…2026年6月26日
#29Zavelinski/claude-code-docs-drift-sync1 skillsdocs-drift-syncUse when code changed and the docs may now be stale, or before merging/releasing when docs must match the code. Triggers - "are the docs still correct", "did my change break the docs", "docs drift", "update the docs for this change", "check docs against…2026年6月26日
#30Zavelinski/claude-code-content-repurpose-engine1 skillscontent-repurpose-engineUse when the user wants to turn ONE source asset (a blog post, transcript, video, or long note) into many platform-native posts. Triggers - "repurpose this", "turn this into posts", "make a thread from this", "content repurposing", "one input many outputs",…2026年6月26日
#31Zavelinski/claude-code-geo-aeo-audit1 skillsgeo-aeo-auditUse when the user wants their content to be CITED by AI answer engines (ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot) rather than ranked in classic search. Triggers - "GEO audit", "AEO", "answer engine optimization", "make this citable by…2026年6月26日
#32Zavelinski/claude-code-skill-security-scan1 skillsskill-security-scanUse BEFORE installing a third-party Claude Code skill or plugin, or to audit skills you already have. Triggers - "is this skill safe", "scan this SKILL.md", "audit this skill before I install it", "check this plugin for prompt injection", "vet this skill",…2026年6月26日
#33Zavelinski/claude-code-adversarial-verify1 skillsadversarial-verifyUse when the user wants proof that a code change actually works, not an opinion. Triggers - "verify this change", "prove this works", "did my fix actually work", "adversarially check this", "are you sure", /adversarial-verify. Re-derives what the change is…2026年6月26日
#34Zavelinski/claude-code-scheduled-sop-runner1 skillsscheduled-sop-runnerUse when the user wants to turn a repeatable task, SOP, or skill into a scheduled routine that runs on its own and only notifies them when something fails or a success condition is not met. Triggers - "run this every morning", "schedule this SOP", "check X…2026年6月26日