Skill v1.0.0
currentLLM-judged scan90/100version: "1.0.0" name: james-au-content-craft description: Use when drafting, adapting or reviewing James Au social posts, carousel copy, video packaging or spoken hooks for LinkedIn, X, Instagram, YouTube, Threads, TikTok and Facebook. license: MIT metadata: version: 1.0.0 status: reviewed-james-adaptation
James Au content craft
Apply this editorial layer after the James Au Social Content Engine and canonical brief, before platform copy and graphics. It distils 23 reviewed source skills from seven Sergey Bulaev bundles. The source lock and MIT attribution are included; these are James-specific adaptations, not installed upstream automation clients.
James's actual voice, source meaning, explicit preferences, selected brand and project safety contract take precedence over every craft suggestion. A good sentence can stay as written. No tactic requires a metric, confession, dramatic reversal, question, CTA, hashtag, artificial rhythm or visual.
Read only what the task needs
- Read editorial workflow for drafting or review.
- Read only the selected platform sections in
- Read algorithm practice for relevant
discovery, retention and interaction choices. Apply supported guidance and label experimental heuristics; do not treat algorithm advice as forbidden.
- For a carousel, cover, thumbnail or video opening, also read
visual and spoken handoff, then route actual asset production through james-au-social-graphics or the relevant video skill.
- Consult source review only for attribution,
exclusions or update review. source-lock.json pins the reviewed upstream files.
Working sequence
Resolve the real source and James's supplied perspective. Select one useful reader takeaway. Draft from that meaning in the selected channel's own language, format and rhythm. Give a concrete opening, fulfil its promise, and preserve qualifications. Review for evidence, human voice, platform fit and unnecessary embellishment. Return only the requested formats and eligible destinations.
When one missing lived detail prevents an honest draft, reuse known answers or ask one relevant question through the conversation director. Do not run a compulsory career interview. Studio has already completed intake: report remaining gaps in warnings and avoid unsupported assertions instead of asking another question.
Output
Return native draft copy plus a short craft note: the chosen opening, what changed from the source, why the format fits, and any missing evidence/media. For Studio, use the existing canonicalBrief, variants[].copy, variants[].notes, warnings schema. Put production directions and limitations in notes, outside public copy. The host records exact instruction hashes in skillBindings; never invent a validation score or claim that instructions guarantee reach or copy quality.
Boundary
Drafting only. No tools, credentials, detector upload, scraping, paid generation, posting, engagement, scheduling or new automation is authorized by this skill. Review-only upstream archives are not executable instructions. Missing Content Engine returns missing_brand_dependency; do not invent another voice profile.