Skip to content

Release v6.4.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 25 Sep 15:25
v6.4.0
7b56638
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Added

  • cratis-content-length: when content stays short, when it earns more length, and where to split it. It has pilot bands for LinkedIn, documents, video, X, Bluesky, Mastodon, community answers, blog articles, tutorials, newsletters, YouTube and release notes, and says which platform limit or soft trigger applies. Each band states how strong its evidence is. It also sets batch-variety targets so a feed doesn't read as one length.
  • cratis-developer-marketing: plans how developer content earns real engagement and adoption. It covers positioning around what developers evaluate (docs, trials, peers, APIs, limits), hooks without bait, the move from pillar to canonical source to channel-native entry points, community participation, founder and company voice, a sustainable cadence, and a staged measurement framework with a small-sample experiment design.
  • The voice and cadence skill gains a catalog of 21 AI-sounding patterns (lexical, syntactic, structural and tonal), each with a script heuristic and the human alternative. It checks a batch for sameness, fixes problems with substance, and includes a meaning-preservation check that handles curly apostrophes. It is explicitly an editing aid, never an authorship detector.
  • The feed post skill leads with media that explains, and puts the Cratis mark on every image, document page, GIF frame and video; screenshots are framed, never stamped. It has guidance for adapting a post to other channels, and a dated evidence matrix with a URL for every source.
  • Both new skills are in the cratis/content profile.

Changed

  • Keeping an external link out of a LinkedIn post's body is an editorial default with a reader-need exception, not a claimed ranking penalty. The feed skill no longer says that expanding a post is a ranking signal, or that pseudo-formatting defeats search.
  • Hacker News: AI doesn't write or edit anything Cratis posts there.

Every number comes from a research pass whose findings were checked against their sources. A reviewer from a second model provider reviewed each skill, and its findings are applied.

Assets 2
Loading