Common Questions

General

What is Unbrowse?

Unbrowse is infrastructure for AI agents that enables them to interact with any website at the network level — 100x faster and more reliably than traditional browser automation.

Think of it as "Google for agent actions" — we index how to do things on websites, not just information about them.

How is this different from browser automation tools?

Traditional automation (Selenium, Puppeteer):

  • Slow (30-45 seconds per action)

  • Unreliable (70-85% success rate)

  • Requires screenshots and visual parsing

  • Breaks when UIs change

  • Expensive compute costs

Unbrowse (network-level):

  • Fast (<1 second per action)

  • Reliable (95%+ success rate)

  • Direct HTTP communication

  • Stable (works even when UIs change)

  • Minimal infrastructure costs

Who is Unbrowse for?

Three main groups:

  1. Indexers - Anyone who browses the web

    • Earn passive income from browsing normally

    • No technical skills required

  2. Developers - Building AI agents

    • Get universal web access via API

    • No need to build custom integrations

  3. Token Holders - Investors

    • Benefit from usage-based burns

    • Value grows with platform adoption

Is this legal?

Yes. Unbrowse operates within legal boundaries:

  • Uses user-provided credentials (authorized access)

  • Respects robots.txt and site policies

  • Similar to browser automation (established legal precedent)

  • Website owners earn 40% of fees (incentive to allow)

Key legal precedents:

  • hiQ Labs v. LinkedIn (web scraping with public data is legal)

  • Users can automate their own accounts

  • Network-level access not prohibited by law

How do you handle websites that block automation?

Multiple strategies:

  1. TLS/HTTP/2 fingerprinting

    • Match real browser signatures

    • Indistinguishable from human browsing

  2. Cloudflare bypass

    • FlareSolverr integration

    • Challenge solving

  3. User credentials

    • Agents use real user accounts

    • Legitimate authenticated access

  4. Website owner cooperation

    • 40% revenue share incentivizes allowing access

    • Better than blocked scrapers

For Indexers

How much can I earn?

Depends on:

  • Quality of abilities (enterprise > consumer)

  • Usage volume (growing with AI adoption)

  • First-mover advantage (early = more)

Realistic ranges:

Casual (10 abilities, consumer focus):

  • $5-50/month initially

  • $100-500/month at scale

Serious (100 abilities, mix of tiers):

  • $50-200/month initially

  • $1,000-5,000/month at scale

Power (500+ abilities, enterprise focus):

  • $200-1,000/month initially

  • $10,000-50,000/month at scale

Plus FDRY appreciation from burns!

Is my browsing data private?

Yes. Zero-knowledge architecture:

  • Passwords encrypted client-side

  • We never see plain-text credentials

  • Pattern extraction only (not personal data)

  • You control what gets captured

  • Exclude sensitive sites

What we capture:

  • Network request patterns (how sites work)

  • API endpoints and structures

  • Response schemas

What we DON'T capture:

  • Screenshots or recordings

  • Keystrokes

  • Plain-text passwords

  • Personal browsing history (anonymized)

What if multiple people index the same site?

Variant system:

  • Multiple variants can exist for same action

  • LAM chooses best variant for each request

  • Quality metrics determine priority:

    • Success rate

    • Speed

    • Error handling

    • Recency

First-mover advantage:

  • Early high-quality variants get most traffic

  • But better variants can compete

  • Incentivizes quality improvement

Can I index on work accounts?

Check your employer's policies first.

Considerations:

  • Company data policies

  • Terms of employment

  • Account ownership

  • Intellectual property

Safe approach:

  • Use personal accounts

  • Index public SaaS tools

  • Avoid proprietary systems

We recommend: Stick to personal subscriptions you pay for.

For Developers

How do I integrate Unbrowse into my agent?

Quick start:

from unbrowse import Client

# Initialize
client = Client(api_key="your_api_key")

# Search for ability
abilities = client.search("post to LinkedIn with image")

# Execute
result = client.execute(
    ability_id=abilities[0].id,
    params={
        "text": "Exciting news!",
        "image_url": "https://...",
        "tags": ["AI", "automation"]
    }
)

print(result.data)

Full docs: For Developers

What's the pricing?

Simple, transparent:

  • Search: $0.0001 (0.01¢) per query

  • Execute: $0.001 (0.1¢) per action

Typical workflow:

  • 1 search + 1 execution = $0.0011 (0.11¢)

Volume discounts: Available for >1M requests/month

Compare to alternatives:

  • Browser automation: $0.01-$0.10 per action (10-100x more expensive)

  • Custom scrapers: $5,000-$50,000 setup (never breaks even)

Are there rate limits?

Default limits:

  • 1,000 searches/hour

  • 100 executions/hour

  • 10,000 searches/day

  • 1,000 executions/day

Upgraded plans:

  • Pro: 10x limits

  • Enterprise: Unlimited (contact us)

Why limits:

  • Prevent abuse

  • Ensure quality of service

  • Fair resource allocation

What if an ability fails?

Automatic handling:

  1. Retry logic (3 attempts with backoff)

  2. Fallback variants (try alternative patterns)

  3. Error reporting (detailed failure info)

Your code:

try:
    result = client.execute(ability_id, params)
except UnbrowseError as e:
    print(f"Error: {e.message}")
    print(f"Suggestions: {e.suggestions}")

Success rate guarantees:

  • 95%+ success rate on most abilities

  • Refund on verified failures

  • Continuous improvement via LAM

Can I create private abilities?

Coming soon:

  • Private ability creation via SDK

  • Custom patterns for internal tools

  • Team-level sharing

  • Usage tracking and analytics

Current workaround:

  • Use indexer extension on internal tools

  • Set privacy level to "private"

  • Only your agents can access

Tokenomics

Why does FDRY have value?

Three value drivers:

  1. Utility - Required for platform access

    • Agents pay in USDC, converted to FDRY

    • Indexers earn in FDRY

    • Platform usage creates demand

  2. Scarcity - Deflationary burns

    • 40% of revenue → buybacks

    • 50% of buybacks → permanent burns

    • Supply decreases daily

  3. Network effects - Growing platform

    • More abilities → better coverage

    • More agents → more usage

    • More usage → more burns

    • More burns → higher value

Not speculation — usage-backed value.

How do the burns work?

Automatic process:

  1. Agent pays $0.001 for execution

  2. 40% ($0.0004) goes to FDRY Treasury

  3. Treasury buys FDRY from open market

  4. 50% of bought FDRY sent to burn address

  5. Tokens permanently removed from circulation

Scale:

  • $1M monthly revenue → $400K to buyback → $200K burned

  • $100M monthly revenue → $40M to buyback → $20M burned

Cumulative effect:

  • Month 1: $73 burned

  • Month 12: $16,335 burned

  • Month 24: $199,650 burned

  • At scale: Millions burned monthly

When should I buy FDRY?

Not financial advice, but consider:

Bull case for early entry:

  • AI agents inevitable (all major companies building)

  • Current solutions failing (70-85% reliability)

  • First-mover network effects

  • Deflationary burns from day one

  • Early holders benefit most from scarcity

Risks:

  • Platform adoption may be slower than projected

  • Competitors may emerge

  • Regulatory changes possible

  • Token price volatile

Investment thesis:

  • Bet on AI agents needing internet access

  • Bet on Unbrowse execution

  • Bet on usage-based burns creating value

Time horizon: 12-24 months for significant appreciation

What's the token distribution?

Total supply: 1,000,000,000 FDRY (1 billion)

Distribution:

  • 40% Community (indexers, early adopters)

  • 30% Treasury (development, partnerships)

  • 20% Team (4-year vesting)

  • 10% Early investors (2-year vesting)

Vesting prevents dumps:

  • Team tokens unlock gradually

  • Investor tokens locked for 2 years

  • Community tokens earned via contribution

Can I stake FDRY?

Coming Q1 2026:

  • Stake FDRY for additional yield

  • Earn from platform revenue share

  • Governance voting power

  • Early access to features

Current:

  • Hold FDRY → benefit from burns

  • Earn FDRY → index abilities

  • Spend FDRY → execute abilities

Technical

What technology does Unbrowse use?

Frontend:

  • Next.js 16 (web interface)

  • React 19 (UI components)

  • Better Auth (authentication)

Backend:

  • Mastra 0.17 (AI agent framework)

  • Node.js 20 / Bun (runtime)

  • PostgreSQL + pgvector (database)

  • Infraxa (vector search)

AI:

  • XAI Grok (ability generation)

  • Google Gemini (embeddings)

  • Custom LAM (Large Action Model)

Execution:

  • curl-impersonate (TLS fingerprinting)

  • FlareSolverr (Cloudflare bypass)

  • E2B (sandboxed code execution)

Payments:

  • X402 (Solana smart contracts)

  • Atomic multi-party splits

Is the code open source?

Partially:

Open source:

  • SDK and client libraries

  • Extension (coming soon)

  • API documentation

  • Smart contracts (audited)

Proprietary:

  • LAM training data

  • Ability generation algorithms

  • Network execution optimizations

Why: Competitive advantage while maintaining transparency where it matters (payments, security).

How do you ensure security?

Multi-layered approach:

  1. Credential encryption

    • Client-side AES-256-GCM

    • Zero-knowledge architecture

    • Local-only decryption

  2. Sandboxed execution

    • E2B isolated environments

    • No access to host system

    • Resource limits enforced

  3. Network security

    • HTTPS/TLS everywhere

    • Certificate pinning

    • DDoS protection (Cloudflare)

  4. Code audits

    • Third-party security reviews

    • Automated vulnerability scanning

    • Bug bounty program (coming soon)

  5. Compliance

    • SOC 2 Type II (in progress)

    • GDPR compliant

    • Regular penetration testing

What happens if Unbrowse goes down?

High availability design:

  • 99.9% uptime SLA

  • Multi-region deployment

  • Automatic failover

  • Database replication

If we go offline:

  • Cached abilities still work

  • Local extension continues capturing

  • Payment processing queued

  • Automatic retry logic

Disaster recovery:

  • Daily backups

  • Point-in-time recovery

  • Ability export API (you own your data)

Can I export my abilities?

Yes! You own your contributions.

Export options:

  1. JSON export - Raw ability definitions

  2. Code export - Executable wrappers

  3. HAR export - Original captures

Access:

unbrowse export --type=json --output=my-abilities.json

Use cases:

  • Backup your work

  • Migrate to self-hosted

  • Custom implementations

Support & Community

Is there a community?

Yes! Join us:

Discord:

  • General chat

  • Indexer tips

  • Developer Q&A

  • Announcements

Twitter/X:

  • Product updates

  • Success stories

  • Technical deep dives

GitHub:

  • SDKs and examples

  • Issue tracking

  • Feature requests

Monthly calls:

  • Community AMAs

  • Product roadmap

  • Technical workshops

How can I provide feedback?

We love feedback!

Feature requests:

  • GitHub issues

  • Discord #feature-requests

  • Email: feedback@unbrowse.ai

Bug reports:

  • GitHub issues (preferred)

  • Discord #bug-reports

  • Email: bugs@unbrowse.ai

General feedback:

  • Discord

  • Twitter/X

  • Email: hello@unbrowse.ai

User research:

  • Participate in user interviews

  • Beta testing programs

  • Early access features

Still have questions? Join our Discord or email lewis@getfoundry.app.

PreviousError HandlingNextTechnical FAQ

Last updated