Skip to content

Anthropic Standards Update

Latest

Choose a tag to compare

@SamaritanOC SamaritanOC released this 18 May 11:22

Structural refactor to align with the Anthropic Agent Skill standard. No change to the rewrite logic; the skill produces the same output on the same input as v1.0.0.

Frontmatter

  • Moved License: MIT out of metadata to a top-level license: MIT key (lowercase, per standard)
  • Added top-level compatibility field, platform-agnostic
  • Added metadata.homepage pointing to https://smbconsultants.ai

Structure

  • Extracted the AI vocabulary list (formerly Section 2A) to references/ai-vocabulary.md and explicitly wired it into SKILL.md so the agent loads it on every humanize pass
  • Added VERSION.md to track release history
  • Added README.md for distribution (excluded from the skill zip)
  • Removed the author and homepage block from the SKILL.md body; that information now lives in metadata

Vocabulary list expansion

Expanded references/ai-vocabulary.md from ~37 entries to ~59, organized into five sections:

  • High-priority replacements (most common tells)
  • Corporate and consultancy tells (9 new entries: leverage, utilize, facilitate, streamline, seamless, holistic, paradigm, synergy, ecosystem)
  • Inspirational and aspirational tells (7 new entries: unlock, empower, resonate, journey, elevate, thrive, embark on)
  • Marketing and lifestyle tells (2 new entries: curated, immerse)
  • Inflation tells (4 new entries: plethora, myriad, paramount, quintessential)
    Also added three new throat-clearing opener patterns to Section 2B: “In an era where…”, “At the heart of…”, “It’s worth noting that…”

Self-consistency fixes

  • Removed all em dashes from SKILL.md body. The skill’s own hard rule is em dash elimination, and the v1.0.0 instructions violated that rule three times.
  • Added an em dash check as item 6 of the Step 3 final review pass