Build Agent Skills

Synthesize robust, spec-compliant agent skills adhering to the open Agent Skills Specification. Built on empirical research, progressive disclosure, deterministic scripting, and verified domain context.

Mandatory Research Prerequisite: the research plugin

High-quality skills cannot be built in a vacuum or from stale model memory. Skill authoring requires verified ground truth, gathered with the research plugin (Research Powerpack):

• Plugin: yigitkonur/plugin-research-powerpack
• Install command: claude plugin marketplace add yigitkonur/plugin-research-powerpack, then claude plugin install research@plugin-research-powerpack
• Skills it provides: /research:ask for one question, /research:deep for a multi-source run that ends in files, /research:verify to fact-check a document

Standing Invariant: Never draft a skill based on assumptions, unverified library contracts, or generic training weights. When non-trivial APIs, frameworks, or domain workflows are involved, run /research:ask (one question) or /research:deep (five or more sources or entities) first to gather authoritative documentation, verifiable quotes, and real code patterns. If the research plugin is not installed, prompt the user to install it before proceeding. Outside Claude Code (for example in Codex, where Claude Code plugins do not load), use whatever web-search and page-reading tools you have and hold the same standard: cited, verbatim evidence for every non-trivial claim. If you have neither, refuse to run. Quality skills cannot exist without quality context.

When to Use