Business content orchestrator

Business-Content Orchestrator

The single entry skill for go-to-market writing and outbound. It locates the task on the lane × audience map and delegates to one of 10 specialist spokes. The cross-cutting model every spoke shares — the source-derived Voice Profile, the research/signal substrate, and the channel contract that keeps a message native to its surface — lives in business-content-core; read it before drafting anything reputation-, launch-, or fundraising-sensitive.

Routing map (intent → spoke)

Establish the voice & evidence (shared substrate — almost always first)

• Reusable style from real source material → brand-voice (produces the VOICE PROFILE every other lane consumes)
• Market sizing, competitor scans, fund/diligence research → market-research (produces the evidence other lanes cite)

Publish (own channels)

• Long-form: articles, guides, tutorials, newsletter issues → article-writing
• Platform-native systems: X / LinkedIn / TikTok / YouTube threads, scripts, calendars, repurposing → content-engine
• Search visibility: technical SEO, on-page, schema, Core Web Vitals, keyword mapping → seo

Market (the launch)

• Multi-channel launch as one orchestrated campaign (positioning → landing copy → email → ads → social → video scripts) → marketing-campaign (itself an orchestration layer; it pulls brand-voice, content-engine, and seo)

Fund (investor-facing)

• Decks, one-pagers, memos, models, accelerator apps that must stay mutually consistent → investor-materials
• Cold emails, warm-intro blurbs, follow-ups, update emails to angels/VCs/accelerators → investor-outreach

Reach (outbound & network)

• Full find → score → warm-path → enrich → draft pipeline for high-value contacts → lead-intelligence
• The standalone graph-ranking engine (bridge math, decay tuning, network-gap analysis) → social-graph-ranker (the model lead-intelligence runs internally; break it out when the user wants the ranking itself)

Folded spokes

Additional specialists folded into this cluster. They cover narrative craft (long-form storytelling, fiction technique) and internal/HR-facing writing — adjacent to the go-to-market lanes above and sharing the same voice-first, evidence-backed discipline. Route to one when the named concern appears; load it on demand exactly like the spokes above.

• Curated quote/aphorism library — theme matching, add/research/search, newsletter variety → aphorisms (reach for the opener/closer quote when drafting a newsletter or article)
• Layered fiction system (Storr science + rhetorical figures) — story bibles, character arcs, chapters, prose → writestory
• Autonomous multi-agent novel pipeline (Writer→Validator→Auditor→Reviser CLI, review gates) → inkos-multi-agent-novel-writing (book-length production at scale; distinct from writestory's single-author craft loop)
• Seven-step novel methodology guide — how to organize a novel project end-to-end → novel-writer-workflow-guide
• Natural dialogue technique — character-consistent, subtext-driven conversation that reveals character and advances plot → natural-dialogue-techniques (pair with writestory/novel-writer-workflow-guide when drafting dialogue scenes)
• Internal communications — 3P updates, leadership/company newsletters, FAQs, incident & status reports in standard company formats → internal-comms

Picked-up spokes

Recently absorbed specialists (from antigravity-awesome-skills). They sit on the build / launch / measure side of the same product story the GTM lanes serve — what you're shipping, how it becomes a business, and whether it works — so the orchestrator can route product-/SaaS-shaped asks, not just writing-shaped ones. Load each on demand exactly like the spokes above.

• Production LLM product engineering — RAG, prompt engineering at scale, structured-output validation, latency/streaming UX, cost control → ai-product (reach for it when "the product itself is AI" and must hold up in production)
• AI-API-wrapper business — defensibility, usage metering, rate limiting, model selection, output quality for a paid wrapper tool → ai-wrapper-product (the business/moat angle on top of ai-product's build angle)
• Indie micro-SaaS launch playbook — validation → MVP → pricing → launch → sustainable revenue for solo/small founders → micro-saas-launcher (bootstrapped GTM; pairs with marketing-campaign for the launch beat)
• Notion-template (digital-product) business — design-to-sell, Gumroad/Lemon Squeezy, marketplaces, support systems → notion-template-business (productizing templates into recurring revenue)
• Senior PM toolkit — 30+ frameworks (RICE/JTBD/Kano/North Star), PRD/roadmap templates, 32 SaaS metrics with formulas → product-manager (structured product analysis: PRDs, roadmaps, prioritization, metrics — feeds positioning into the publish/fund lanes)
• Segment CDP engineering — Analytics.js, server-side tracking, Protocols tracking plans, identity resolution, governance → segment-cdp (the measurement substrate behind launches; route here when "instrument the product / capture clean event data" comes up)

Standard Operating Flow

1. Locate the task: which lane (publish / market / fund / reach) and which audience (public, investor, prospect).
2. If the output is voice-sensitive, run brand-voice first and reuse its VOICE PROFILE — never re-derive style per draft or default to generic AI copy. If it makes a business claim, ground it via market-research. (Model in business-content-core.)
3. Delegate to the spoke(s). Multi-step asks fan out in lane order — e.g. "launch this" → brand-voice → marketing-campaign (which pulls content-engine + seo); "raise a round" → investor-materials → investor-outreach; "find and reach buyers" → social-graph-ranker → lead-intelligence.
4. Return: chosen spoke(s), the voice/evidence inputs they consumed, the target channel(s), and the next action (draft, deck, sequence, or outreach list).

Guardrails

See business-content-core. In short: voice-first, evidence-backed, channel-native — one VOICE PROFILE flows everywhere instead of per-draft mimicry; every business claim traces to a real source, not invention; each message is shaped for its surface (the same copy reused across email, LinkedIn, and X is a tell). Never auto-send outbound — draft and let the user approve. The cluster's value is content that reads like a specific person with receipts; don't quietly let it collapse into platform slop.

Loading spokes on demand

To keep CLI startup context lean, this cluster's spokes are not separately registered as skills — only this orchestrator and its *-core are enumerated. When you route to a spoke named above, load it on demand by reading its file:

~/.agents/skill-clusters/skills/<spoke-name>/SKILL.md (or skills/<spoke-name>/SKILL.md inside the skill-clusters repo).