You don't need to write clever prompts. The system already knows your business. Your job is to be clear about what you need.
Every task you hand off follows the same format. It doesn't need to be long. It needs to be clear.
What I need: Follow-up email to Sarah Chen at Greenleaf. We met today to discuss their Q3 content calendar.
Context: She wants to move from monthly to biweekly blog posts. Budget is fine, her concern is quality at higher volume. I told her we'd send a revised scope by Friday.
Voice / format: Warm, she's been a client for 2 years. Short. Confirm the scope deadline and recap the volume conversation.
Save to: Outputs/Comms/greenleaf-followup.docx
What I need: Prep doc for my call with David Park at Summit Financial tomorrow at 2pm.
Context: This is our second meeting. First call was an intro, he's interested in quarterly reporting automation. His team is 8 people, mostly analysts.
Voice / format: Internal doc for me. Talking points, what to ask, potential concerns to address.
Save to: Outputs/Meetings/summit-prep.md
What I need: Weekly status summary across my active projects.
Context: I'm managing Greenleaf (content calendar), Summit (proposal stage), and Ridgeline (brand refresh in progress).
Voice / format: Internal. Structured by client. Progress, risks, next actions for each.
Save to: Outputs/Comms/weekly-status.md
Each specialist reads your context files before responding. The output sounds like your company wrote it.
Drafts, reviews, and improves any message going to or from a client — status updates, follow-ups, meeting summaries, difficult conversations, scope discussions. Writes in your voice, calibrates tone to the situation, treats every message as relationship capital.
Writes brand-aligned content across every format — emails, proposals, blog posts, LinkedIn, case studies, newsletters, website copy. Adapts tone to the channel while staying inside your company's established voice. A LinkedIn post sounds different from a proposal, but both sound like you.
Reviews any draft for brand consistency — voice, tone, messaging, visual identity. Every judgment traces back to your company's communication standards. Flags what's off, explains which principle it violates, shows the corrected version. Sharpens your voice; doesn't replace it.
Tracks project status, surfaces risks before they become problems, maps dependencies across people and timelines, keeps everyone aligned. Produces structured status reports — Progress / Risks / Next Actions — where every action has one named owner and every status is honest.
Conducts structured research and synthesizes findings into actionable summaries — prospect briefs, competitive analysis, market overviews, topic backgrounders. Organizes by theme, evaluates source quality, rates confidence on key claims, and always tells you what's missing.
Takes complex, lengthy, or messy information and turns it into sharp summaries for senior audiences — leadership teams, partners, clients, board members. Bottom line up front, always. Readable in under three minutes. Four formats: executive summary, briefing doc, condensed version, one-pager.
Makes sense of numbers — raw data, spreadsheets, metrics, reports — and turns them into clear findings for non-technical audiences. Leads with what matters, highlights outliers, explains what the data means for the work at hand. Always flags when data is incomplete.
Structures ideas into clear, persuasive slide narratives. Decides what goes on each slide, what order makes the story land, how to adapt the narrative for the audience. One idea per slide. No generic titles. Three narrative frameworks mapped to audience type — client pitch, internal update, executive briefing.
Handles the numbers side of engagements — fee proposals, cost estimates, budget tracking, profitability analysis, pricing comparisons. Clear tables, honest assumptions, flagged risks. States every assumption. Uses round numbers where precision doesn't matter. Never hides bad news.
Maps, documents, and improves how work gets done — operational workflows, recurring processes, handoff points. Produces usable SOPs in plain imperative language, process maps with owners, improvement recommendations graded by effort. Works from what exists before suggesting changes.
These handle multi-step tasks in one pass and keep the system's memory current. The value isn't any one piece — it's the orchestration and continuity across everything you do.
Produces a complete pre-meeting brief so you walk in informed, prepared, and able to send a follow-up within five minutes of hanging up. Works for both prospect meetings and existing-client check-ins.
A consolidated end-of-week brief across every active engagement. Run it once on Friday and walk away with a complete picture plus ready-to-send client check-ins. No engagement falls through the cracks.
Runs as a scheduled task at the end of each workday, or manually when you say "wrap up the day." Reviews everything from the session and writes a structured entry — what got done, what was decided and why, what files changed, what's still open, what to pick up tomorrow. Next morning, Claude reads this first, so every session starts where yesterday left off.
Makes your voice profile more accurate over time by studying real evidence. Paste examples of what Claude wrote versus what you actually sent — the skill extracts the patterns that define your real voice and updates your profile so every skill produces better output going forward.
Schedule daily-log within each project: Cowork → Scheduled → set to run at end of day. Important: each project needs its own schedule — one schedule doesn't run across all projects.
Done isn't a chat response you copy-paste. Done is output you can open, review once, and use.
Not a response in the chat window. Not bullet points. A file you can open, read once, and use. If you're reading Claude's output in the chat and copy-pasting it into a document, your brief is missing the "Save to" line.
If you're rewriting more than 20% of the output, something is off. Either the context files need updating (your voice has drifted), or the brief was too vague. Both are fixable.
The system gets better as you use it. When you correct something, tell Claude why. If it keeps getting your tone wrong on a specific type of communication, run /refine-voice with a real example of what you would have written. The system learns.
If you're rewriting more than 20% of the output, either your context files need updating or your brief was too vague. Both are fixable.
If you're new to the team or new to Cowork, here's your path to first real output.
Open Claude Desktop, go to Settings → Plugins, and search for Nobo Foundation. Install it. This gives you the full skill set — specialists, workflows, onboarding, everything in this guide.
/onboard to create your profile.This creates your personal profile — your role, your voice, how you prefer to work. It's a conversation, not a questionnaire.
/new-project.Loads your company context, your personal profile, and asks a few questions about the specific engagement. Every specialist now knows the landscape.
Pick something you actually need done today. Use the four-line brief. Save it to your Outputs folder. Review what comes back.
Check that /new-project ran in this project folder. If it sounds almost right but not quite, run /refine-voice with a before/after example of what you would have written.
A weekly dispatch from the field. What's working, what isn't, what I'm building next.
friday field notes →