← ClaudeAtlas

revenantworks-foundation-skillwrightlisted

Builds, audits and packages Agent Skills, with their evals. Trigger to build, audit (security too), score or package a skill or SKILL.md; to design or integrate a pack; whether a skill beats incumbents; for trigger evals, test cases or assertion suites (skill, prompt card, agent spec); to slim a skill; when a skill did not fire or a run cost too much; for a prose pass on a skill's own files; to port or sanitize a skill set; to build or contract-test the MCP tools a plugin or mod ships; or say /rw-skillwright (evals, slim, diagnose, mcp, refresh, port, pack, integrate, upkeep). Prompt tuning is promptwright's; running a suite, claude plugin eval's; config slims, rigwright's; other docs, commscribe's; brands, brandscribe's; runtime grants, gatewarden's.
revenantworks/claude-skills · ★ 2 · AI & Automation · score 81
Install: claude install-skill revenantworks/claude-skills
# revenantworks-foundation-skillwright *history in CHANGELOG.md · sources and parity register in SOURCES.md · Apache-2.0 (LICENSE)* Turn a one-line intent into a shipped, install-ready Agent Skill, or port an existing set to a new owner or purpose. A new skill is researched against current best practices, built to beat its strongest incumbents, tested, and packaged; an existing skill gets the same standards as an audit. **Builds spec-clean neutral:** a member carries its pack's structural identity (name segments + frontmatter token) and no applied styling, brand or voice. **Build workflow:** Intent → Pack & profile → Research → Parity verdict → Design catalog *(one gate)* → Build → Self-audit → Package Ships no executable code; tools and fallbacks (no file tools → in-chat file content; optional `zip` and stdlib `python3` skip cleanly, neither needed for a build) are in `compatibility`. The rule it builds by: **no undeclared dependencies** — any tool, script, or sibling a built skill needs is named in its frontmatter and docs. ## Turn shape 1. **One catalog, one gate, no drip-feed.** Every decision set (design catalog, audit findings) is shown complete, once, with a recommendation per item. One approval round follows; "apply all" / "just build it" anywhere in the request skips the gate. Never re-open a settled catalog with unsolicited additions. 2. **Gates render by the tool-list test.** Before writing a gate, scan the tool list: if any tool presents tappable options or