Claude-skill-registry hud

Project dashboard that guides you through setup and tracks progress

install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/hud" ~/.claude/skills/majiayu000-claude-skill-registry-hud && rm -rf "$T"
manifest: skills/data/hud/SKILL.md
source content

Project HUD

Shows status and actively guides you through what's next.

Core Principle

Vision emerges from doing, not declaring.

Don't force users through setup forms. Let them build. Capture context into the knowledge docs AS you work together - the docs become a record, not a gate.

If they tell you what they want to build → get right into it, save context in background. If they're lost → ask "What are you building?" and go from there.

Usage

/hud                    # Full dashboard + guided next step
/hud --compact          # One-line status

Workflow

Step 1: Get Context via MCP Tools (ONE CALL)

A. Get unified context:

Call: mcp__jfl-context__context_get
Parameters: { "taskType": "general", "maxItems": 30 }

This returns EVERYTHING in one call:

  • Recent journal entries (what happened across sessions)
  • Knowledge docs (vision, roadmap, narrative, thesis, tasks)
  • Code file headers (@purpose tags)

DO NOT read individual markdown files. The context MCP tool aggregates everything.

B. Pull CRM pipeline:

./crm list

Look for these statuses that need attention:

  • 🟠
    IN_CONVO
    - Active conversations, may need follow-up
  • 🟡
    REACHED_OUT
    - Waiting for response
  • 🔴
    HOT
    - Urgent action needed
  • 📞
    CALL_SCHEDULED
    - Prep needed

C. Assess state from context_get response:

  • Journal section → recent work, decisions, what's in progress
  • Knowledge section → vision status (EMERGENT vs DECLARED), ship date from roadmap
  • Code section → what files exist, their purposes

D. If you need to search for something specific:

Call: mcp__jfl-context__context_search
Parameters: { "query": "your search terms" }

Step 2: Route Based on State

IF foundation docs are templates/empty:
  → ONBOARDING MODE (guide through setup)

IF foundation is done but no brand:
  → BRAND MODE (guide to /brand-architect)

IF everything set up:
  → EXECUTION MODE (show status, next tasks)

Step 3A: New Project (Foundation Empty)

Don't force setup. Get them building.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
{PROJECT NAME}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

What are you building? Let's get into it.

(Vision emerges from doing, not declaring -
I'll capture context as we go.)

When they tell you:

  • Start building immediately - use skills, write code, whatever they need
  • Capture context in background - save what they said to VISION.md
  • As you make decisions together, record them in appropriate docs
  • Don't interrupt flow to "fill out forms"

Step 3B: Brand Mode (Foundation Done, No Brand)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
{PROJECT NAME}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Vision: ✓
Roadmap: ✓
Narrative: ✓

Next up: Brand identity.

Ready to create your visual identity?
I'll generate logo marks, colors, and typography.

Say "let's do it" or /brand-architect

Step 3C: Execution Mode (Everything Set Up)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
{PROJECT NAME}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Ship: {date} ({days} days)
Phase: {current phase}
Memory: {X} docs, {Y} memories indexed

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
PIPELINE (from ./crm list)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🟠 Jack (Nascent) - IN_CONVO - needs follow-up docs
🟠 Wes (Formation) - IN_CONVO - waiting on beta timing
🟡 Avi (Coinfund) - REACHED_OUT - no response yet

{Only show IN_CONVO, REACHED_OUT, HOT - skip COLD}

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
RECENT WORK
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
{from journal + memory query - recent decisions/work}
{what's next based on that context}

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
THIS WEEK
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
{from knowledge/TASKS.md or inferred priorities}
1. {most urgent based on pipeline + context}
2. {second priority}
3. {third priority}

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
NEXT ACTION
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
{Pick ONE specific thing based on:
 - Urgent pipeline items (follow-ups due)
 - Recent journal entries (what's in progress)
 - This week's priorities from TASKS.md}

What do you want to tackle?

Compact Mode (--compact)

{project} | {days}d | Phase: {phase} | Next: {action}

Key Behaviors

  1. Never end with open questions like "What do you want to work on?"

    • Instead: Suggest the specific next action
    • Or: Ask a specific question to move forward
  2. Detect returning users

    • Check journal for recent work (last few entries)
    • Query memory for "recent decisions" or "current focus"
    • "You were working on X. Want to continue?"
  3. Guide, don't report

    • Bad: "Your docs are templates. Fill them in."
    • Good: "What are you building? Tell me in 2-3 sentences."
  4. One thing at a time

    • Don't overwhelm with all missing pieces
    • Focus on the immediate next step

Dependencies

  • Works with minimal setup (just CLAUDE.md)
  • Better with
    knowledge/
    docs populated
  • User context from
    suggestions/{name}.md
  • Context Hub MCP is the hub - use
    mcp__jfl-context__context_get
    for unified context
  • Journal files at
    .jfl/journal/<session-id>.jsonl
  • CRM via
    ./crm
    CLI (Google Sheets backend)

MCP Tools Available

ToolPurpose
mcp__jfl-context__context_get
Get unified context (journal + knowledge + code)
mcp__jfl-context__context_search
Search across all context sources
mcp__jfl-context__context_status
Check Context Hub status
mcp__jfl-context__context_sessions
See activity from other sessions
mcp__jfl-memory__memory_search
Semantic search in memory
mcp__jfl-memory__memory_add
Add a memory directly