TCEA Skill Suite · Claude-Powered · Nonprofit Edition

Seven skills that write,
design, diagram, and
plan for mission-driven orgs.

Purpose-built AI skills for nonprofits and education-adjacent organizations — craft donor communications, build board decks, design impact infographics, map workflows, generate org charts, evaluate programs, and architect volunteer training. Powered by Claude.1

7
Skills
8
Cowork Steps
Claude
Powered

Seven skills. One workflow.

Each skill is a precision-engineered instruction set directing Claude to produce professional, mission-ready output — no prompt guesswork required.

Blog Writer

Impact stories, donor updates, and advocacy posts in a consistent nonprofit voice.

Content · Nonprofits

PPTX Builder

Download-ready board and funder decks via PptxGenJS — branded, with speaker notes.

Presentations · .pptx

Infographics

Blueprint-first workflow rendering publication-ready PNG impact visuals from program data.

Visualization · PNG

Mermaid Designer

Grant intake, case management, and volunteer onboarding flows with TCEA color theming.

Diagrams · Code

Draw.io XML

Importable mxGraph XML for org structures, service delivery maps, and program flowcharts.

Diagrams · XML

Rubric Builder

Grant proposal, program evaluation, and volunteer performance rubrics — observable and measurable.

Evaluation · Nonprofits

Training Architect

Structured volunteer and staff training session plans — specific enough for any facilitator to deliver.

Training · Capacity Building

How to use Claude

Eight surfaces — each with a specific purpose, a nonprofit-ready first prompt, and common mistakes to avoid.2

Feature What it does & first prompt Pro tip Common mistake
💬 Chat
The starting point for most users.
"Rewrite this donor appeal so it leads with impact, not need."
Pro tip: Turn on Extended Thinking + Search before every prompt — two clicks that change everything.
Mistake: Pasting your mission statement into every new chat. That's what Projects are for.
Cowork
Reads your files and creates real documents — Word, Excel, PDF — directly into your folder.
"Read my files first. Then ask me questions before you start."
Pro tip: Write one .md file about your org — mission, tone, key programs. Claude stops producing generic language.
Mistake: Dumping your entire grants folder. 5 focused files beat 50 unorganized ones.
📁 Projects
Save instructions and context files once. Every chat inside that Project inherits them.
Create a "Grant Writing" project → upload 3 awarded proposals → every new draft matches your voice.
Pro tip: One Project per recurring task type — grant writing, donor comms, board reports. Don't build one mega-Project.
Mistake: Uploading 30 reference docs. You curate the best — not the AI.
🧩 Artifacts
Claude builds interactive files you can use, edit, and download inside the chat.
"Build a volunteer tracking dashboard with hours by program area — totals update in real time."
Pro tip: Ask for changes after it builds. Artifacts are live — iterate inside the same chat.
Mistake: Thinking Artifacts are just demos. Ask for what you'd normally build in a spreadsheet or Canva.
📊 Excel
An add-in that reads your actual formulas, tabs, and cell references — not flattened screenshots.
"Why is my budget vs. actual formula showing #REF? Trace the error."
Pro tip: Install → Insert → Get Add-ins → search "Claude by Anthropic." Open with Ctrl+Alt+C.
Mistake: Expecting it to automate button clicks. It builds, cleans, and explains — not VBA macros.
🔗 Connectors
Link Google Drive, Notion, Slack, and 50+ tools. Claude searches them mid-conversation.
"Find our Q3 program report in Drive" — no uploading, no screenshots.
Pro tip: Use the Gamma connector in Cowork to go prompt → outline → finished funder presentation slides.
Mistake: Thinking it syncs live. Claude searches your tools on demand — it doesn't watch them 24/7.
🔌 Plugins
One-click skill packs that add slash commands for Communications, Data, Legal, and more.
Install Marketing plugin → type /draft-post → get a LinkedIn post announcing your annual appeal.
Pro tip: Type / in any chat to see every slash command available.
Mistake: Installing all 11 plugins. Each adds context Claude must process. Pick 2 that match your actual work.
📚 Skills
Reusable instruction packs that make Claude better at specific tasks — automatically.
Settings → enable Code Execution → browse Skills library → install one.
Pro tip: Write your own Skill.md — org tone rules, grant writing checklist, impact reporting format.
Mistake: Confusing Skills with Projects. Projects hold your files. Skills teach Claude how to do a task.

Blog Writer

Write impact stories, donor updates, and advocacy posts for nonprofit audiences — conversational, structured, and mission-aligned.

  • Opens with a human story, pivots to the systemic issue, then to the org's response
  • Structured format: opening scenario, pivot, H2 sections, data table, call to action
  • Readability rules built in — short paragraphs, active voice, no jargon
  • Tool roundup rules when applicable: name specific platforms, note cost and access
  • Banned-words list eliminates filler phrases and AI-sounding language automatically
Download blog-writer.skill
Nonprofit Impact Post Draft
ProgramReachFY
Food Access2,400 families2024
Job Training318 graduates2024
Youth Mentorship540 youth2024

PPTX Builder

Professional board and funder decks built with PptxGenJS — download-ready .pptx files, never just outlines.

  • 16×9 layouts with title, divider, and content slide types — all visually consistent with org branding
  • Speaker notes auto-included so board presentations work without a live presenter
  • First prompt: "Build a 10-slide board deck showing Q3 program outcomes and budget variance."
Download pptx-builder.skill

Infographic Designer

Converts program data or impact narratives into a structured blueprint, then renders publication-ready PNG infographics for annual reports and grant submissions.

  • Blueprint-first: hierarchy, layout, and text confirmed before rendering begins
  • Bar charts, icon grids, flow diagrams, comparison tables, and radial charts
  • Splits complex topics into 2–3 panels automatically when needed
  • All text clean, typo-free, and optimized for print and digital distribution
Download infographic-designer.skill
Blueprint → PNG · Annual Report
Volunteer Hours
12,400
9,200
6,600
Donor Split
Major
Mid
Annual
Workflow
Data
Blueprint
Render PNG

Mermaid Designer

Generates branded Mermaid diagram code — grant intake flows, case management sequences, volunteer onboarding paths, and program logic models — with TCEA color theming applied automatically.

  • Selects the right diagram type from context: flow, sequence, ER, Gantt, state, mindmap
  • Three built-in themes: Navy/Gold (TCEA brand), Slate/Teal (neutral), Light/Accessible (print)
  • Node, edge, and subgraph style rules applied consistently — no manual formatting
  • Copy-paste-ready for GitHub, Notion, Confluence, or mermaid.live
Download mermaid-designer.skill
flowchart TD · Grant Intake Flow
Application Received
Eligible?
Program Review
Decline & Refer

Draw.io XML Builder

Produces complete, importable mxGraph XML for draw.io and diagrams.net — org structures, service delivery networks, and program logic maps ready to open and edit.

  • Full mxGraphModel XML output — paste into draw.io via Extras → Edit Diagram
  • Color palette locked to TCEA brand values — no arbitrary hex codes
  • Layout grid and node sizing rules ensure readable, consistent diagrams every time
  • Recipes for org charts, service maps, program logic models, and swimlane workflows
Download drawio-xml-builder.skill
draw.io | FileExtrasView ← Edit Diagram
Executive Dir.
Programs
Development
Volunteers
Grants Mgr

Rubric Builder

Builds analytic, holistic, and single-point rubrics for grant proposal scoring, program evaluation, and volunteer performance review — observable and measurable every time.

  • Three formats: Analytic (criteria × levels), Holistic (level paragraphs), Single-Point (mastery column)
  • Performance indicators matched to the evaluation's purpose and cognitive level
  • Enforces observable language — no vague terms like "good," "appropriate," or "understands"
  • Header block included: evaluation name, type, total points, and scoring notes
Download rubric-builder.skill
Grant Proposal Scoring Rubric
Program Officer Review · 20 pts · Analytic
Criterion 4Exceeds 3Meets 2Approaching
Need StatementData-driven, local, specific population namedClear need; some local dataNeed present but broad or anecdotal
Logic ModelActivities, outputs, outcomes clearly linkedOutcomes tied to activitiesOutcomes vague or missing
Evaluation PlanNamed measures, timeline, and data owner2+ measures; collection noted1 measure; no collection plan

Training Session Architect

Complete volunteer and staff training plans for single sessions or full onboarding sequences — specific enough for any facilitator to deliver without additional guidance.

  • Performance verbs matched to objectives — no vague "understand" goals
  • Five-phase sequence: Hook, Direct Instruction, Guided Practice, Independent Work, Closure
  • Differentiation built in: scaffolding for new volunteers, extension for experienced staff
  • Unit plan variant: learning pathway, essential questions, and session sequence table
Download lesson-plan-architect.skill
Volunteer Intake & Safety Training
New Volunteers · 60 min
0–8 min
Hook / Activation
Share a client story (anonymized). Ask: "What would you need to feel prepared for this?"
8–22 min
Direct Instruction
Model client intake process using a sample scenario. Check for understanding at step 3.
22–40 min
Guided Practice
Pairs complete a mock intake together. Facilitator circulates and addresses questions.
50–60 min
Closure
Exit ticket: "Name one thing you'll do differently in your first shift based on today."

8 Steps to Set Up Your Entire Claude Cowork

Do these once. They run forever. Requires Claude Pro ($20/mo) and the macOS or Windows desktop app.3

1

Download The App

Go to claude.com/download. You need a Pro plan. macOS or Windows only. Open the app.

2 minutes. Don't overthink it.

2

Click The Cowork Tab

Open the app. Click Cowork at the top. Select Opus 4.6. Turn on Extended Thinking.

Wrong model = bad output.

3

Build This Exact Folder

Create one folder: "Claude-Work." Inside it, 4 subfolders:

  • ABOUT ME — your org, mission, and how you write
  • PROJECTS — one subfolder per active grant or initiative
  • TEMPLATES — awarded proposals, strong reports
  • OUTPUTS — where Claude saves finished documents

Cowork has read/write access. Keep it tight.

4

Create 3 Context Files

Inside ABOUT ME, create:

  • about-org.md — mission, programs, population served
  • our-voice.md — communication tone, 2–3 writing samples
  • our-rules.md — ask before starting, show a plan, never delete

One great file beats 50 random uploads.

5

Set Instructions

Settings → Cowork → Edit Instructions. Paste:

"I'm [Name] at [Org], [Role]. Read my files before every task. Ask questions before executing. Show a plan. Never delete without approval."

Set once. Runs forever.

6

Stop Prompting. Start Clicking.

Paste this for every task:

"I want to [TASK]. Read all files first. Ask me questions using AskUserQuestion before you execute. If something is off, generate new questions. Do not guess."

Claude forces you to be clear — you don't have to be.

7

Install Your First Plugin

Customize → Browse Plugins → Install. Type / to see slash commands.

  • Marketing → /marketing:draft-content
  • Data → /data:explore
  • Legal → "Review this MOU."

Start with 2.

8

Connect Your Tools

Settings → Connectors → Browse → Add. Google Drive, Notion, Slack. 50+ more. Free on all plans.

Claude reads your tools mid-conversation. No copy-pasting.

Claude stops being a chatbot. Starts being a coworker.

Make Claude Ask YOU First

Stop writing perfect prompts. Use this 3-step pattern to get outputs built on your exact context — not generic assumptions.4

Step 1

Start with any task

Template for every nonprofit task:

I want to [TASK] so that [SUCCESS]. First, read uploaded files. DO NOT start yet. Ask me clarifying questions (use AskUserQuestion) to refine the approach. Only begin once we've aligned.

Example: "I want to write a foundation appeal letter so that we secure $25K for our food access program."

Step 2

Claude creates a form

Claude generates an interactive AskUserQuestion form with clickable options. No typing required. Example questions it might ask:

  • What foundation is this for?
  • What program area does the grant cover?
  • Do you have a prior relationship with this funder?
  • Do you have an awarded proposal to use as a template?
Step 3

Answer the questions

Claude processes your answers, confirms a plan, and only then begins writing. The output reflects your org, your funder relationship, and your program — not a generic template.

"Perfect — I have everything I need. Here's the plan…"

Why this works4

You stop writing perfect prompts — Claude forces clarity instead.

Iterate through questions, not through editing bad outputs.

With 1M+ token context, hallucinations drop significantly.5

Claude generates clickable buttons. You click. It executes.

When to reset mid-conversation

If a grant draft or donor letter is going in the wrong direction, paste this to reset without losing context.

Works in Cowork and standard Chat.

Copy–paste this
We are getting sidetracked. We need to start over without losing context. First, generate an AskUserQuestion form so we decide what went wrong and where we must go instead. From my answer, generate a second AskUserQuestion form to complete the task.

Reusable capabilities for Claude agents

Install with a single command. Bundle slash commands, agents, hooks, and MCP servers into shareable packages that work across Claude Code, Cowork, and Claude.ai.6

Claude skills are reusable capabilities for AI agents, easily installed with a single command. They bundle slash commands, agents, hooks, and MCP servers into shareable packages that work across Claude Code, Cowork & Claude AI. Nonprofit teams can install skills that pull live documentation, manage long coding sessions, or add persistent memory — without re-explaining context every session.
Skill

context7

Pulls version-specific documentation from source repos directly into Claude's context. Claude gets live, accurate API docs — no outdated training data. Useful for nonprofits building custom tools or integrations.7

github.com/upstash/context7
Skill

ralph-loop

Turns Claude into a self-driving development agent. Implements a stop-hook pattern for long-running, multi-hour autonomous coding sessions — resets context between tasks automatically.8

github.com/anthropics/claude-plugins-official
Skill

claude-mem

Adds long-term memory to Claude so it carries context and preferences across sessions. No more re-explaining your org's programs, funders, or voice every conversation.9

github.com/anthropics/claude-mem
Skill · 96K+ installs

frontend-design

The most popular Claude skill. Transforms Claude's UI output from generic defaults into distinctive, production-grade interfaces. Ideal for building donor portals or volunteer dashboards.10

claude.com/plugins
Skill

code-review-agents

Automated PR review using multiple parallel specialized agents — comment analysis, test coverage, silent failure detection, confidence scoring. Useful for lean tech teams.11

github.com/anthropics/claude-code/tree/main/skills

Anthropic Courses

Free official training from Anthropic — from basic fluency to building with the Claude API. All courses available at anthropic.skilljar.com.12

📘

Claude 101

Learn to use Claude for everyday work tasks, understand core features, and explore resources for more advanced learning.

anthropic.skilljar.com
💻

Claude Code in Action

Integrate Claude Code into your development workflow and build autonomous coding agents.

anthropic.skilljar.com
🔧

Building with the Claude API

The full spectrum of working with Claude's API — from basic calls to advanced agentic patterns.

anthropic.skilljar.com
🧩

Introduction to MCP

Build Model Context Protocol servers and clients from scratch using Python.

anthropic.skilljar.com
🎓

AI Fluency for Students

Develop AI fluency skills that enhance learning and career planning.

anthropic.skilljar.com
🏫

AI Fluency for Educators

Empowers faculty, instructional designers, and educational leaders to apply AI.

anthropic.skilljar.com
🌐

AI Fluency: Framework & Foundations

Learn to collaborate with AI systems effectively, efficiently, and ethically.

anthropic.skilljar.com
⚙️

Introduction to Agent Skills

Build, configure, and share Skills in Claude Code — reusable markdown instruction sets for agents.

anthropic.skilljar.com

How TCEA skills work

Add a skill file to your Claude Project's knowledge base. From that point forward, Claude reads it automatically and applies its rules to every matching request in that project.

1
Upload the skill

Add the .skill file to your Claude Project. One upload covers every future conversation in that project.

2
Describe your task

Natural trigger phrases — "write a donor update about…", "draw a flowchart of our intake process…", "build a rubric for evaluating…" — activate the right skill automatically.

3
Get a finished file

Receive a complete deliverable — blog draft, .pptx deck, PNG infographic, Mermaid diagram, XML, rubric, or training plan — ready to use immediately.

Ready to use all seven?

Upload each skill file to a Claude Project and start producing professional content, diagrams, and evaluations in minutes.

View All Skills How It Works

Sources & References

  1. 1TCEA AI Skills Suite — original site content. TCEA, 2024–2025.
  2. 2"How to use Claude" feature overview including Chat, Cowork, Projects, Artifacts, Excel, Connectors, Plugins, and Skills. Original graphic content, circulated 2025. See also: docs.claude.ai
  3. 3"8 Steps to Set Up Your Entire Claude Cowork." Instructional content from circulated social media guide, 2025. Desktop app available at claude.com/download
  4. 4"Make Claude Ask YOU First." AskUserQuestion method guide, including Steps 1–3 and "Why This Works" rationale. Circulated social media content, 2025.
  5. 5"With 1M+ token context, hallucinations drop." Claim from source guide content, 2025. Anthropic's extended context window documentation: docs.anthropic.com
  6. 6"Claude skills are reusable capabilities for AI agents." Official Anthropic description. Skills library: github.com/anthropics/claude-code/tree/main/skills
  7. 7context7 MCP skill. Upstash, MIT License. github.com/upstash/context7
  8. 8ralph-loop (Ralph for Claude Code). Autonomous development loop with intelligent exit detection. github.com/frankbria/ralph-claude-code
  9. 9claude-mem. Persistent memory compression for Claude Code. github.com/anthropics/claude-mem
  10. 10frontend-design skill — 96K+ installs as reported at claude.com/plugins. Anthropic, 2025.
  11. 11code-review-agents (Code Review Plugin). Anthropic official skills repository. github.com/anthropics/claude-code/tree/main/skills
  12. 12Anthropic official course catalog. Courses include Claude 101, Claude Code in Action, Building with the Claude API, Introduction to MCP, AI Fluency series, and Introduction to Agent Skills. anthropic.skilljar.com