Bloom Discovery Skill

Don't browse 13,000 skills. Browse use cases.

---

The Problem

Without Bloom	With Bloom
Browse 13,000 skills manually	Describe your intent, get matched use cases
Install skills one by one, hope they work together	Each use case = tested skill combination
No idea if your setup is complete	Verify config → see what's missing
No proof of your expertise	Claim SBT → on-chain verified identity
Recommendations based on popularity	Recommendations based on how you actually work

---

What It Does

1. Personality Analysis (MentalOS)

Reads your USER.md and conversation history to map your builder personality:

• 5 Types: Visionary, Explorer, Cultivator, Optimizer, Innovator
• 4 Spectrums: Learning (Try First ↔ Study First), Decision (Gut ↔ Analytical), Novelty (Early Adopter ↔ Proven First), Risk (All In ↔ Measured)
• Hidden Patterns: 6 detectors surface insights you might not see yourself
• AI-Era Playbook: Personalized leverage, blind spots, and next moves

/bloom

~60 seconds. No surveys. No auth.

2. Use Case Discovery

Intent-driven skill browsing. You say what you want to accomplish; Bloom matches you to curated use cases — each a tested combination of skills.

• Personality-aware ranking (novelty seekers → new use cases, risk-averse → skip DeFi)
• Keyword frequency threshold (≥ 3 mentions) filters noise
• Each use case comes with skill requirements and verification status

"find use cases for me"

3. Verify Configuration & Claim SBT

Scans your installed skills (~/.openclaw/skills/) and verifies your setup against use case requirements.

• Match percentage + missing capabilities
• Claim flow: verified config → mint SBT (on-chain proof)
• No wallet? Fallback to web link at bloomprotocol.ai
• SBT = soulbound token proving your verified configuration

"verify my config"
"claim my SBT"

4. Usage Metrics (Opt-in)

Anonymized reporting to improve recommendations for everyone.

• Reports only: skill name + usage frequency
• No conversation content. No personal data.
• Opt-in only — never sends data without your explicit consent

---

Quick Start

OpenClaw

clawhub install bloom-discovery
/bloom

Developers

git clone https://github.com/bloomprotocol/bloom-discovery-skill.git
cd bloom-identity-skill
npm install
cp .env.example .env

# Run from session file
npx tsx scripts/run-from-session.ts \
  ~/.openclaw/agents/main/sessions/<SessionId>.jsonl \
  <userId>

---

MentalOS Spectrum

Spectrum	Low (0)	High (100)	Question
Learning	Try First	Study First	Jump in or read the docs?
Decision	Gut	Analytical	Instinct or methodical?
Novelty	Early Adopter	Proven First	Bleeding edge or battle-tested?
Risk	All In	Measured	Go big or hedge?

5 Personality Types

Type	Style
The Visionary	High conviction + high intuition — backs bold ideas early
The Explorer	Low conviction + high intuition — experiments widely
The Cultivator	High contribution — builds communities and shares knowledge
The Optimizer	High conviction + low intuition — doubles down on what works
The Innovator	Low conviction + low intuition — spots patterns others miss

---

Security & Privacy

✅ Local analysis — Conversation text analyzed on your machine, never uploaded ✅ Local Differential Privacy (ε=1.0) — Spectrum scores noised via Laplace mechanism ✅ SHA-256 fingerprint — Conversation hashed locally; only irreversible hash stored ✅ Minimal transmission — Server receives personality type + approximate scores only ✅ Minimal writes — Only writes ~/.bloom/agent-id.json (returning user token, 0600 perms) and bloom-discoveries.md. Never modifies session files or USER.md ✅ Atomic file writes — Uses tmp+rename pattern to prevent corruption ✅ Input validation — Regex guards on session IDs, Twitter handles, file paths, agent IDs ✅ No shell injection — All external commands use execFile (no shell interpolation) ✅ Timeouts — All network requests have 5–10s abort signals ✅ Opt-in metrics — Usage data never sent without consent ✅ Open source — Audit the algorithm yourself

❌ Raw conversation text never sent to any server ❌ Wallet private keys never transmitted ❌ No PII collection ❌ No background data collection ❌ File reads restricted to user's home directory

---

Configuration

# Required
JWT_SECRET=your_secret_key_here

# Recommended
BLOOM_API_URL=https://api.bloomprotocol.ai
DASHBOARD_URL=https://bloomprotocol.ai
NETWORK=base-mainnet

See .env.example for all options.

---

Architecture

Conversation + USER.md
        │
        ▼
  ┌─────────────┐
  │  Personality │ ← Local analysis (LDP ε=1.0)
  │  Analyzer    │
  └──────┬──────┘
         │
    ┌────┴────┐
    ▼         ▼
┌────────┐ ┌────────────┐
│Use Case│ │Skill Recomm.│
│Matcher │ │  Pipeline   │
└───┬────┘ └──────┬─────┘
    │              │
    ▼              ▼
┌────────┐  ┌───────────┐
│Verify  │  │ Dashboard │
│& Claim │  │   Card    │
└────────┘  └───────────┘

---

Links

• Homepage: bloomprotocol.ai
• ClawHub: clawhub.ai/unicornbloom/bloom-discovery
• Source: github.com/bloomprotocol/bloom-discovery-skill
• Dashboard: bloomprotocol.ai/agents
• Docs: bloomprotocol.ai/docs

---

Built by Bloom Protocol

Agent-native intent economy curation platform.

Built with @openclaw, @coinbase, and @base