A blank Claude Project is a blank chat with extra steps.A configured one is an expert that already knows your business.
Claude Projects are persistent workspaces — custom instructions, a knowledge base, persistent context across sessions. The capability is in the box. The hard part is designing the instructions that hold up over months of daily use, naming knowledge files so RAG actually retrieves them, and writing starter prompts that compound the setup instead of fighting it.
The Claude Project Blueprint Library is ten drop-in blueprints — one per business function. Sales, founder, support, marketing, ops, finance, HR, product, legal, content. Each blueprint is custom instructions + a named knowledge-base map + 10–12 starter prompts + pro moves. Plus a universal quickstart and 12 months of updates. One-time $79.
Most homegrown Projects work for a week, then degrade.
Claude Projects are the right primitive for “Claude that knows my business” — persistent custom instructions, a knowledge base, context that carries across sessions. The Anthropic Help Center documents the capability clearly and the setup is fast.
The gap shows up at month two. Instructions that started clean get bloated as people add edge-case handling. Knowledge files get named ad-hoc, so the instructions can’t reference them reliably (RAG retrieval is only as good as your file hygiene — misnamed files don’t retrieve). Starter prompts drift toward generic “help me with sales” rather than the specific patterns the Project is configured for. The Project still answers, but it answers like a normal chat with extra steps.
The blueprints solve the design problem. Each one is built to hold up under daily use — focused instructions that don’t bloat, named files that retrieve cleanly, starter prompts that compound the setup, pro-moves patterns that get sharper every week you use them.
Homegrown Projects start with a clean role definition and accumulate edge-case handling over time. By month two, the instructions are a 1,500-word grab-bag and Claude's behavior gets less consistent, not more. The blueprints stay focused on what actually changes behavior.
One 80-page “everything” doc retrieves worse than eight focused, well-named files. The blueprints' Knowledge-Base Map names files exactly the way the instructions reference them — so RAG retrieval actually fires on the right source.
The fastest way to discover what a Project is actually good at is to fire a curated starter library — 10-12 prompts that exercise the main workflows. Each blueprint ships that set; you skip the “what should I ask it” phase entirely.
Clear about the lane. No inflated promises.
- Ten production-grade Claude Project blueprints — one per business function (sales, founder, support, marketing, ops, finance, HR, product, legal, content) — plus a universal quickstart guide.
- Each blueprint = custom instructions + named knowledge-base map + 10-12 starter prompts + setup & pro-moves section.
- Designed to hold up under daily use — focused instructions, clean retrieval, prompts that compound.
- Cited foundation on the May 2026 Anthropic Help Center docs for Projects.
- 12 months of updates and on-demand additions (refinements to the ten blueprints, vertical variants, and new blueprints as catalog demand surfaces them).
- Pairs with the Skills Packs (loaded into the Project as workflow capabilities) and the SOP Engine (documents the human process).
- A replacement for the free Anthropic Help Center docs on Projects — those are the foundation.
- A skills pack. Skills are workflow capabilities loaded INTO a Project; blueprints ARE the Project setup.
- An MCP server. Connect MCP servers separately for live CRM / Slack / GitHub data feeds.
- A fine-tuned model or custom Claude variant. Standard Claude Pro / Max / Team / Enterprise.
- A subscription. One-time $79 with 12 months of updates and on-demand additions.
- A guarantee of clean RAG retrieval if you ignore the naming convention. File hygiene is the user's job; the kit teaches it.
Ten blueprints. One per business function.
Each blueprint is independent. Stand up one in five minutes; add the next when you’re ready. The starter prompt library for each is the fastest way to discover what the Project is actually good at — and the cleanest shop pattern is one Project per function, not one Project per person.
Turn Claude into your sales operations partner — pipeline hygiene, discovery prep, follow-up that actually gets sent, honest deal review.
Founder-led sales · SMB sales teams · anyone running a pipeline in a CRM and a head
Discovery prep, follow-up after a call, cold outreach on stage, objection on the spot, deal review, pipeline triage, proposal skeleton, competitor situation, re-engage a dead lead, weekly forecast narrative
A chief-of-staff-grade thinking partner — strategy pressure-testing, board and investor updates, exec memos, and the weekly “what actually matters” triage.
Founders · solo operators · small leadership teams running the whole company out of their head
Weekly priority triage, board update draft, investor monthly update, strategic memo pressure-test, exec one-pager, hiring decision framework, weekly retrospective, board-question prep
Make a support team faster and more consistent — triage and tag tickets, draft on-brand replies, write knowledge-base articles, QA responses before they go out.
SMB support teams · SaaS customer-support functions · dev teams running their own support
Ticket triage + tag, on-brand customer reply, KB article draft, escalation decision, reply QA, holding response for unknown issue, churn-risk flag, internal status note
Turn Claude into your marketing team — briefs that don't waste the work, campaign plans, on-channel copy, and repurposing that multiplies one asset into ten. Grounded in your proof; never invents a stat.
Founders running their own marketing · small marketing teams · agencies standardizing client work
Campaign brief from a rough idea, per-channel copy variants, headline/hook ladder, content calendar build, repurpose long-form to short, weekly performance narrative, on-voice rewrite, claim audit against proof
Turn Claude into your operations partner — SOPs that stay current, process audits that find the leaks, and vendor/project tracking that doesn't live in your head.
Ops leads · COOs and chiefs-of-staff · founders whose business runs on tribal knowledge
SOP from a loose description, process audit, vendor scorecard, project status roll-up, weekly ops review, runbook for a recurring incident, onboarding checklist, ops decision memo
Turn Claude into your finance partner — reporting narratives, scenario modeling, and board-ready numbers explained in plain language. Works your numbers; never invents them.
Founders who own the finances · fractional CFOs · finance leads prepping for boards and investors
Monthly close commentary, runway scenarios, board pack narrative, cohort math walk-through, variance explanation, cash-flow forecast story, investor-question pre-mortem, KPI narrative
Turn Claude into your people partner — job descriptions, structured interview kits, policy first-drafts, and performance reviews. Fair, structured, and consistent — not legal advice.
Founders and managers who hire and run reviews without an HR team · people-ops leads standardizing the basics
JD from a rough role spec, structured interview kit, lawful-question check, scorecard build, performance-review draft, policy first-draft, candidate comms, offer-letter rationale
Turn Claude into your product partner — PRDs and specs, release notes, and feedback synthesis that turns scattered input into a clear signal. Grounded in real user feedback; never invented.
Founders doubling as PM · product managers · technical teams standardizing product docs
PRD from a loose idea, feedback synthesis from raw notes, release notes, sprint summary, prioritization frame, post-launch debrief, weekly product narrative, customer-interview prep
Turn Claude into your drafting partner for the paperwork that piles up — first-draft agreements, policies, and admin documents from your own templates. Drafts only; does NOT give legal advice — verify with counsel.
Founders and ops/admin leads who generate routine documents and want fast, consistent first drafts to send to counsel
First-draft agreement from your template, policy first-draft, NDA from template, admin-doc drafting, change-of-terms notice, vendor-form pre-fill, contract-review summary, redline assist for counsel review
Turn Claude into your editorial team — long-form drafted in your real voice, an editorial pipeline that doesn't stall, and a content engine that stays on-brand and honest.
Founders who publish · content marketers · creators and editorial teams running a regular cadence
Long-form draft on a topic, editorial calendar build, headline ladder, repurpose to short-form, brief from a rough idea, on-voice rewrite, evergreen update, weekly publishing roll-up
Four parts. Designed to compound.
Every blueprint in the library uses the same four-part structure so you can stand up the second and third Project as fast as the first.
The system prompt for the Project. Defines Claude's role, operating rules, output defaults, and guardrails for that function. Drop into the Project's instructions field with a handful of bracketed fields filled in (company, ICP, tone, escalation rules). About 400-700 words of focused setup.
The exact list of files to upload — what goes in each, priority order (must-have / high / medium), and the naming convention. The instructions reference these files by name, so naming hygiene is part of the kit. Misnamed files don't retrieve.
10-12 ready-to-paste prompts that exercise the Project's main workflows. Each prompt is calibrated to the Project's instructions and knowledge base — they're not generic prompts that happen to mention the function; they're prompts that compound the Project's specific setup.
The 5-minute build path and the patterns that compound the Project's value over time. The “close the loop” pattern (feed new objections / patterns back into the knowledge files), the “one Project per motion not per rep” rule, and the cross-attach points for MCP servers, Skills Packs, and the SOP Engine.
From folder to first useful answer in five minutes.
Four steps. You do them once per blueprint. After that, every session inside the Project starts from the configured expert instead of the blank chat.
Claude.ai → Projects → Create Project. Name it after the function (Sales Engine, Founder's Office, Support Hub). Add a one-line description.
Open the matching blueprint file. Copy the Custom Instructions block, paste into the Project's instructions field, replace every {BRACED FIELD} with your real values, save.
Work through the blueprint's Knowledge-Base Map checklist. Upload the document types it lists. Name files exactly as the map suggests — the instructions reference these names.
Start a new chat inside the Project. Paste any prompt from the Starter Prompt Library. Output is now grounded in your business, not a generic chatbot reply. Five minutes from create to first useful answer.
The one discipline that makes this work: file-naming hygiene. Name files by content, not date. One topic per file. Match the names in the instructions exactly. RAG retrieval is only as good as your file hygiene — and the blueprints are built around the named files, not generic blob storage.
The integrity moat.
Exactly what you get for $79, and what you don’t.
- Ten production-grade Claude Project blueprints — Sales Engine, Founder's Office, Support Hub, Marketing Studio, Operations Desk, Finance Cockpit, People & HR, Product Cockpit, Legal & Admin, Content Room.
- Universal quickstart guide for first-time Project builders.
- 12 months of updates (refinements as Anthropic Projects evolves) and on-demand additions (vertical variants, new blueprints driven by catalog demand).
- File-naming convention and KB hygiene guidance — the discipline that makes RAG retrieval actually work.
- Starter prompt libraries (10-12 prompts per blueprint).
- Setup + pro-moves guidance for each blueprint, including cross-attach points for MCP servers and Skills Packs.
- A Claude subscription. You need Claude Pro / Max / Team / Enterprise separately.
- Skills (workflow capabilities). Pair with the Solo Founder / Agency Operators / LST / Ecommerce Skills Packs.
- MCP servers (live system integrations). Pair with the Agent Orchestration Cookbook for that work.
- Custom blueprint authoring on your behalf. The kit teaches the pattern; you adapt it to bespoke functions.
- A managed service. You own the Project setup and the file hygiene.
- Legal advice (Legal & Admin produces FIRST DRAFTS only; verify with counsel before relying on anything legally binding).
Pairs naturally with the Solo Founder Skills Pack ($79) and the Agency Operators Skills Pack ($89) — Skills are workflow capabilities loaded into the Project; the blueprints are the Project setup that hosts them. The cleanest shop pattern: one Project per function, each scaffolded by a blueprint, with the matching Skills Pack layered on top.
For shops also documenting human process, pair with the AI-First SOP Engine ($79) — the SOP Engine documents the human process; the blueprint puts an AI partner inside the Project that runs alongside it.
For teams adding live system integrations, pair with the Agent Orchestration Cookbook ($79) — MCP server patterns connect HubSpot / Salesforce / Slack / GitHub data feeds directly into the blueprint’s starter prompts.
The questions operators actually ask before standing up their first Project.
Four parts. (1) Custom instructions — the system prompt that defines Claude's role, operating rules, output defaults, and guardrails for that function. Drop it into the Project's instructions field with a few bracketed fields filled in. (2) A knowledge-base map — the list of named files to upload, with what goes in each and a must-have / high / medium priority order. The instructions reference these files by exact name, so naming hygiene is part of the kit. (3) A starter prompt library of 10-12 ready-to-paste prompts that exercise the Project's main workflows. (4) A setup + pro-moves section with the 5-minute build path and the patterns that compound the Project's value over time.
All ten ship today — the full planned set. Sales Engine (pipeline, discovery, follow-up, deal review, forecast). Founder's Office (strategy pressure-testing, board and investor updates, exec memos). Support Hub (triage, on-brand replies, KB articles, reply QA). Marketing Studio (briefs, campaign plans, on-channel copy, repurposing). Operations Desk (SOPs, process audits, vendor/project tracking). Finance Cockpit (reporting narratives, scenario modeling, board-ready numbers). People & HR (job descriptions, structured interview kits, policy drafts, performance reviews). Product Cockpit (PRDs, release notes, feedback synthesis). Legal & Admin (first-draft agreements and policies from your templates — not legal advice). Content Room (long-form in your voice, editorial pipeline, on-brand cadence). A universal quickstart ships alongside, plus 12 months of updates and on-demand additions.
Pro is enough to use every blueprint at full strength. Projects are available on Pro, Max, Team, and Enterprise per the May 2026 Anthropic Help Center documentation. Team and Enterprise add sharing — one owner builds the Project once, every teammate fires the same starter prompts against the same vetted knowledge. The blueprints work identically on Pro for solo operators; Team plans amplify the consistency benefit when multiple people use the same Project.
You could absolutely write your own — and the Anthropic Help Center has the foundational guidance for free. The library's IP is the operating-grade design: which custom-instructions patterns hold up over months of daily use, which knowledge files compound the Project's value vs which ones add noise, the exact file-naming convention that makes RAG retrieval actually work (the instructions reference files by exact name; misnamed files don't retrieve), and the starter prompt set that surfaces what the Project is actually good at vs what it isn't. Most homegrown Projects work for a week then degrade; the blueprints are designed to compound.
Yes — they live at different layers. Skills are loaded into a Project as workflow capabilities (the Solo Founder pack ships six skills; the Agency Operators pack ships six more). Blueprints are the Project itself — the role, the rules, the knowledge base, and the starter prompt set. The cleanest setup at most shops: one Project per function (Sales / Founder / Support), each with the matching blueprint as the scaffold, and Skills layered on top for specific workflows. The SOP Engine ($79) and the Blueprint Library complement each other — the SOP Engine documents the human process; the Blueprint Library puts an AI partner inside the Project that runs alongside the process.
Not to use the kit — only to use a Project for your specific work, and only to the extent you choose. The knowledge-base map names file types (Pricing & Terms, Objection Library, Help Articles, etc.) — you decide what level of detail to put in each. For sensitive content (named customer accounts, deal financials, employee data), the standard advice applies: redact or summarize before upload, and confirm your Claude plan's data handling matches your policy. Projects on Pro/Max do not train Claude models on your inputs per the May 2026 Anthropic data policy; verify against current Anthropic documentation before uploading regulated data (HIPAA, financial PII, etc.).
All ten planned blueprints now ship. The 12 months of updates covers (1) refinements to the existing blueprints as Anthropic Projects evolves (custom instructions, knowledge-base patterns, starter prompt sets), (2) on-demand additions for verticals or sub-functions catalog buyers request (e.g. a hospitality-specific Operations Desk variant, a creator-economy Content Room variant), and (3) any new blueprint drops that catalog demand surfaces. Cadence is demand-driven, not on a fixed publishing calendar.
30-day no-questions refund. Stand up any one of the ten blueprints against a real workflow you run weekly. If it doesn't save you at least an hour per week — or surface at least one structural improvement to how you'd run that workflow without Claude — email and we refund. Refunds across the catalog are countable on one hand; for Project blueprints specifically, the value is usually obvious by the third use.
Stop running blank chats.
Start running configured Projects.
Ten production-grade blueprints, a universal quickstart, twelve months of updates and on-demand additions. Five minutes from folder to first useful answer. Pair with the Skills Packs for workflow capabilities and the SOP Engine for the human process.
Sold by RedHub AI LLC · Secured by Stripe · redhub.ai