ttm-humanize
Mandatory final-step humanizer. Detects and rewrites AI writing patterns (inflated symbolism, em-dash overuse, rule-of-three, AI vocab, passive voice, filler phrases) and matches the user's voice from BRAND.md samples. Runs on every audience-facing asset before write. Also runnab
What it does
/ttm-humanize
Read and follow the workflow at ${CLAUDE_PLUGIN_ROOT}/workflows/lifecycle/humanize.md.
Required reading:
${CLAUDE_PLUGIN_ROOT}/references/humanizer-patterns.md-- pattern catalog..taketomarket/BRAND.md-- voice + tone + banned words..taketomarket/POSITIONING.md-- must-not-say.
Invocation modes
Mode A -- Auto (called by other skills):
Other producing skills call /ttm-humanize with the draft path as input. The skill rewrites in place and returns control.
Mode B -- Manual:
User runs /ttm-humanize <file> or pastes text. Skill rewrites and either writes back or prints the result.
Next steps
Standard footer: see ${CLAUDE_PLUGIN_ROOT}/templates/next-step-footer.md.
Capabilities
Install
Quality
deterministic score 0.46 from registry signals: · indexed on github topic:agent-skills · 13 github stars · SKILL.md body (726 chars)