Goaly

A goal-oriented operating system for running a one-person business with Claude Code. 9 skills that keep you oriented around goals, KPIs, and the actions that actually move them.

Everything is markdown files. No database required. No SaaS subscription. Just folders of .md files with YAML frontmatter and Claude Code skills that reason over them.

Quick Start

git clone https://github.com/Danm72/goaly.git
cd goaly

Then in Claude Code:

/goaly-setup

Answer 5 questions about your goals, KPIs, clients, and working style. The wizard generates all your starter files. Then:

/mission

Get your first scorecard, task ranking, and spear-sharpening check. No Notion, no API keys, no setup beyond this.

What You Get

10 Skills

Skill	What It Does	Trigger
/mission	Weekly planning (Monday) + daily pulse check	"plan my week", "quick check"
/goaly-triage	Surface actionable items from calendar + email	"anything I'm missing?"
/goaly-coaching-prep	Structured agenda for coaching sessions	"prep for coaching"
/goaly-review-meeting	Transcript → action items → tasks → follow-up	"review notes from [meeting]"
/goaly-meeting-prep	Build agenda from client history	"prep for [client] meeting"
/goaly-client-email	Multi-channel client comms (email, Slack, WhatsApp)	"email [client]"
/goaly-screen-lead	Deep research on inbound leads	"screen this lead"
/goaly-onboard-client	End-to-end: CRM, contacts, contract, project	"new client [name]"
/goaly-ceo-review	Strategic scope challenge before building	"review this plan"
/goaly-setup	First-run wizard — generates all starter files	"set up goaly"

Plus linear-product-ops for Linear board management.

The Mission

The flagship skill. Every Monday it:

1. Pulls data — KPIs, tasks, calendar, email, git activity (all in parallel)
2. Builds a scorecard — traffic-light KPI table (green/yellow/red)
3. Detects patterns — portfolio concentration, stalled tasks, killed-mammoth syndrome
4. Budgets energy — deep work slots available vs committed
5. Runs a retro — what shipped, what was planned but untouched, meta-KPIs
6. Ranks tasks — impact scoring (max 10 points):

Criteria	Points
Moves MRR	+3
Ships product toward paying users	+3
Grows audience/subscribers	+2
Has a hard deadline	+1
Someone external is waiting	+1

Top 3-5 tasks become the Pareto 20%.

7. The Spear-Sharpening Check — if 3+ tasks shipped but zero KPIs moved, it flags: "You're sharpening the spear instead of hunting the mammoth."

Weekday pulse mode: just the scorecard, flags, and today's tasks. 2-3 minutes, zero interaction.

Architecture

Everything Is Markdown

notion-mirror/
├── tasks/           # Business actions (YAML frontmatter + markdown body)
├── goals/           # Strategic goals
├── kpis/            # Measurable outcomes with targets
├── projects/        # Bodies of work
├── clients/         # Client CRM data
├── contacts/        # People
├── interactions/    # Meeting notes, emails, calls
├── brainstorms/     # Ideas and exploration
└── personal-tasks/  # Personal actions

Each file has YAML frontmatter with properties and a markdown body. See notion-mirror/README.md for the full format.

Notion is optional. You can:

• Write files by hand
• Generate them with /goaly-setup
• Sync from Notion with tools/notion-sync
• Export from any tool that outputs markdown

Three-Phase Execution

Every skill follows the same pattern:

1. Collect — gather all data in parallel (no serial API calls)
2. Analyze — reason over collected data (no further I/O)
3. Interact — present findings and ask structured questions

Shared Conventions

All skills read _shared/conventions.md on startup:

• Local-first search (grep markdown files before hitting APIs)
• Tables not prose (scannable output, structured choices)
• Semantic deduplication (grep before creating)
• Run logging (every skill appends to a run log)
• Session learning (corrections saved to feedback.log, read back next session)

Memory System

.claude-memory/ provides persistent context across sessions:

• MEMORY.md — concise index loaded every session
• owner-profile.md — your working style, strengths, preferences
• revenue-strategy.md — pricing, pipeline, diversification rules
• lead-screening.md — red/green flags for vetting leads
• ikigai.md — 4-dimension alignment filter for tasks and opportunities

Optional Tools

The core system works with just markdown files and Claude Code. These tools add capabilities:

Tool	Purpose	Required?
tools/notion-sync	Two-way sync between markdown and Notion	No
tools/email-sync	Sync Gmail threads to local markdown	No
tools/granola-sync	Sync meeting transcripts to local markdown	No
gog CLI	Gmail and Google Calendar access	No — check manually
qmd	Semantic search across markdown files	No — use grep
Granola	Meeting transcripts	No — paste transcripts manually

Customization

Modify Scoring

Edit .claude/skills/goaly-mission/references/scoring-rules.md to change how tasks are ranked. The default weights MRR and product shipping highest.

Add Your Own Skills

Follow the pattern in .claude/skills/:

.claude/skills/your-skill/
├── SKILL.md       # Instructions and flow
├── gotchas.md     # Failure modes (mandatory)
├── references/    # Supporting data
└── templates/     # Output formats

Run bash tests/validate-skill-structure.sh .claude/skills/your-skill/ to verify structure.

Change Entity Types

Edit CLAUDE.md to modify:

• Default properties for tasks, goals, KPIs, projects
• Impact classification rules
• Project routing table
• SMART enforcement criteria

Testing

# Validate all skill structures
for d in .claude/skills/goaly-*/; do bash tests/validate-skill-structure.sh "$d"; done

# Run computation tests for a specific skill
bash tests/mission/test-computations.sh

# Run LLM eval (requires promptfoo)
promptfoo eval -c tests/mission/promptfooconfig.yaml

Project Structure

goaly/
├── CLAUDE.md                    # The operating system config
├── .claude/skills/              # 10 skills + shared conventions
├── .claude-memory/              # Persistent memory templates
├── notion-mirror/               # Your business data (markdown files)
├── email-mirror/                # Optional: synced email threads
├── granola-mirror/              # Optional: meeting transcripts
├── dashboards/                  # Generated mission dashboards
├── tools/                       # Notion, email, and Granola sync
├── tests/                       # Skill tests and fixtures
├── clients/templates/           # Contract and meeting templates
├── docs/                        # Reference documentation
└── scripts/                     # Utility scripts

License

MIT

Credits

Built by Dan Malone. Read the full story: Every Monday I Type /mission — 9 Claude Code Skills That Run My Business.