From wireword
Review LLM agent control words for confusable names. Use for route names, tool aliases, prompt macros, service labels, environment labels, profile names, and other short tokens that control agent behavior.
How this skill is triggered — by the user, by Claude, or both
Slash command
/wireword:wirewordThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Wireword reviews agent control words.
Wireword reviews agent control words.
A control word is any short label that changes what an agent does, where work is routed, which tool runs, which environment is targeted, or which prompt macro is expanded.
Flag labels when they have:
A1 / AI / Aldocs.api / doc.apiProd / prodFACTCHECK_API / FACT_CHECK_APIprod / production / liveA, B, Cplan / plannerUse this format:
HIGH docs.api / doc.api
Reason: edit distance 1. One typo can become another valid route.
Fix: rename to ROUTE_DOCS_REVIEW and ROUTE_DOCS_PUBLISH.
Prefer labels with distinct words:
ROUTE_DOCS_REVIEW
ROUTE_BILLING_EXPORT
ENV_PRODUCTION
ENV_STAGING
MACRO_FACTCHECK_API
Avoid compact labels that rely on one character:
A1
AI
Al
P1
PI
Pl
Telegraph codebooks used similar thinking. The two-letter-difference principle made it less likely that one damaged codeword would silently become another valid codeword. Wireword applies that principle to agent control words.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub brfid/wireword --plugin wireword