Market Brief5 code-layer auditsPre-revenue honestyLaunch + trade + marketplaceAgent API already exposed

The rails are real. The pitch should stay there.

Avax Radar already has live launch, trading, creator payouts and headless API surfaces. The honest wedge is wallet-controlled capital tooling, not sci-fi autonomy claims.

What the product can claim today

The credible sell today is straightforward: Avax Radar already ties launch, discovery, trading and creator revenue together on Avalanche. What is not yet credible as a hard claim is backend-owned user autonomy, fully clean data across every page, and a polished public SDK.

Open jury flowOpen Agent APIOpen discovery
Three hard takeaways
Revenue today is much more likely to come from API + marketplace usage than from the launch fee.
Three hard takeaways
The defensible edge is no-UI operation for external agents, not human curation.
Three hard takeaways
If the copy promises backend-owned autonomy right now, it outruns the code.
Hardening

What changed before pitch

The audit turned into product hardening, not only strategy copy. These changes make the demo, deck and video safer to defend because the riskiest public surfaces no longer rely on fabricated discovery data or open private-wallet reads.

SurfaceBeforeNowNext
Discovery projectsWhen the DB was empty, /api/launches converted legacy in-memory launches into project-shaped records with generated scores./api/launches returns DB records or an honest empty list. On-chain launches stay in /api/chain/launches instead of being disguised as indexed projects.Backfill confirmed chain launches into indexed projects only when real metadata exists.
Project writesPOST /api/launches could write indexed project metadata without a signed wallet.POST /api/launches requires wallet auth, rejects founder mismatch and attaches the authenticated founder user.Expose the founder wallet in the UI as provenance and add duplicate checks by curve/token address.
Vault privacyGET /api/chain/vault/:wallet exposed private platform ledger history to any caller.Vault balance/history requires wallet auth and only returns data for the signed wallet. Vault sync is auth-locked and range-limited.Keep public portfolio reads on-chain only; keep custodial ledger reads private.
Proof receiptsProof used a fixed six-agent count even when the live registry should be the source.Proof reads the live system-agent registry count and keeps provider truth visible, including template mode when no LLM key is active.Separate public proof receipts from private operating metrics if real volume grows.
Positioning

Avax Radar vs pump.fun

The nearest mass-market reference is pump.fun, but the product strategy is not the same. pump.fun wins on zero-friction issuance and speculative volume. Avax Radar is stronger when the story is launch plus post-launch analysis, execution and monetized intelligence.

CategoryAvax Radarpump.fun
Primary jobLaunch, analyze, trade, monetize and operate projects through one wallet-native capital surface.Launch a token instantly and maximize speculative trading throughput.
Who gets paidPlatform earns from x402 calls and marketplace take-rate; creators earn from paid skills and copied strategies.Protocol earns from launch and trading fees; creator upside depends mostly on token attention and downstream volume.
After launchProjects stay in discovery, can be analyzed headlessly, traded, sold into, copied and monetized as intelligence.The core value is the launch and immediate market. Post-launch tooling is not the center of the product.
Operator modelWallet-controlled and headless: external operators sign their own actions while the platform exposes launch, trading and monetization rails.Very low-friction issuance and trading. Intelligence, post-launch operations and monetized analysis are not the center of the product.
Best pitch angleWallet-controlled capital tooling for Avalanche, especially for external agents operating without the UI.The fastest way to create and trade a token.
Phase 1

Five code-layer audits

These are the five layers requested in the strategic audit prompt, reduced to the code-backed conclusion and the narrowest relevant evidence.

1. Differentiated technical layer
Intermediate

Launch creation, buy/sell flows and curve migration are real on Avalanche, but user-owned agents are not backend-autonomous on the server.

You can honestly demo launch, buy and sell. You cannot honestly claim that every connected wallet is already a server-side autonomous agent.

Code evidence
Launch payload reads factory fee and builds real calldata
/Users/maiconguimaraes/shark-avax/backend/src/services/avalanche.ts:122
Chain launch endpoint returns wallet-signed payloads
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:1978
Wallet agents return null for user wallets without private keys
/Users/maiconguimaraes/shark-avax/backend/src/agents/registry.ts:292
System agents are server-owned random wallets
/Users/maiconguimaraes/shark-avax/backend/src/agents/registry.ts:309
2. External infra layer
Intermediate

Avalanche RPC, chain health, on-chain launch reads and signed trading payloads are real. Legacy in-memory launch and treasury paths still coexist.

The live chain rails exist, but parts of the app can still fall back into demo-state behavior that weakens trust if left unexplained.

Code evidence
Chain health checks live block number and factory launch fee
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:1943
Pure on-chain and DB-backed launch reads both exist
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:2041
Legacy createTokenLaunch still fabricates addresses in memory
/Users/maiconguimaraes/shark-avax/backend/src/services/avalanche.ts:996
Treasury fee schedule is USDC-native and the post-curve DEX path now points to Pangolin first
/Users/maiconguimaraes/shark-avax/backend/src/services/treasury.ts:16
3. API + persistence
Intermediate

Postgres-backed projects, chain launches, vault ledger, creator balances and purchases are real. The highest-risk public write/read paths are now auth-locked, but wallet auth is still EVM-first.

This moves the product from open demo posture toward paid operator posture without pretending it is enterprise-hard yet.

Code evidence
Wallet auth verifies EVM signatures only
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:63
POST /launches now requires wallet auth
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:1015
Vault balance/history is now scoped to the signed wallet
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:2265
Marketplace stats come from revenue_log + creator_balances
/Users/maiconguimaraes/shark-avax/backend/src/services/marketplace.ts:544
4. Distributed package / SDK surface
Intermediate

The reusable surface today is HTTP + MCP, not a polished publishable SDK or a consistent external package.

That is enough to sell headless access now, but it means the shortest revenue path is API usage, not developer distribution through npm packages.

Code evidence
Root workspace remains private and app-first
/Users/maiconguimaraes/shark-avax/package.json:1
Backend package is private and build is typecheck only
/Users/maiconguimaraes/shark-avax/backend/package.json:1
Contracts package is also private
/Users/maiconguimaraes/shark-avax/contracts/hardhat/package.json:1
Identity cards still emit did:sol while payment/network are Avalanche
/Users/maiconguimaraes/shark-avax/backend/src/services/agent-identity.ts:23
5. External data integrations
Intermediate

Marketplace revenue, chain reads and recent launch data are real. Discovery no longer fabricates project records from memory, but brainstorm and marketplace listings still include heuristic or seeded surfaces.

Judges can trust some dashboards, but not every narrative surface equally. This page should say that plainly instead of hiding it.

Code evidence
Launches now return DB records without avalanche-memory fallback
/Users/maiconguimaraes/shark-avax/backend/src/routes/api.ts:948
Brainstorm still hardcodes sector/team/github defaults
/Users/maiconguimaraes/shark-avax/backend/src/services/auto-brainstorm.ts:89
Proof reads live agent count and still exposes template mode if no LLM key is active
/Users/maiconguimaraes/shark-avax/backend/src/services/proof.ts:184
Skill marketplace seeds curated treasury listings on first boot
/Users/maiconguimaraes/shark-avax/backend/src/services/marketplace-listings.ts:38
Phase 2

Maturity map

The product is neither fake nor production-clean. It has a real core, mixed data surfaces, and unfinished automation claims.

Launch + trade rails        ██████████░░  real
Creator payouts             ██████████░░  real
Headless agent API          ████████░░░░  intermediate
Discovery + proof feeds     ██████░░░░░░  intermediate
Wallet-controlled agents    ███████░░░░░  intermediate
Public SDK / package        ███░░░░░░░░░  intermediate
Launch + trade rails
10/12
Real C-Chain launch, buy and sell flows already exist.
Creator payouts
9/12
Marketplace payments and creator balances are persisted and split 95/5.
Headless agent API
8/12
The routes are there; the distribution surface is HTTP/MCP, not a polished SDK.
Discovery + proof feeds
6/12
Real data is mixed with seeded and heuristic content.
Wallet-controlled agents
7/12
External wallets can operate headlessly today, but the backend does not own or impersonate user wallets.
Public SDK / package
3/12
The integration surface exists over HTTP/MCP, but there is no stable client library or packaging discipline yet.
Real
Avalanche launch payloads and receipt confirmation
Buy / sell execution rails
Marketplace payments, creator balances and honest leaderboard inputs
Intermediate
Headless Agent API
DB + chain discovery
Vault indexing and off-chain capital accounting
Wallet-controlled external agent operation
HTTP + MCP integration surface
Mock
Pure data feeds across every judge-facing surface
Phase 3

Market and revenue reality

This section uses the current code-level pricing rails and current market comps. The numbers are intentionally conservative.

Unit economics
UnitPriceCostMarginWhy it matters
Launch feeVariable on-chain; contract example shows $0.50-equivalent stablecoinUser signs and pays the transaction gasHigh on paper, too small to matter at current pricingLaunch fee is technically real but not a venture-scale business by itself.
x402 / MCP call$0.50 per call~$0.02-$0.05 for a 6-agent LLM pass plus trivial infra~90%+ gross if usage is realThis is the cleanest software-like monetization rail already in code.
Marketplace copy / skill sale$0.50-$50 ticket range today; platform keeps 5%Treasury payout gas + DB write + support overheadHealthy above ~$2 tickets, weak on tiny ticketsThis is the best current path for creator revenue and platform take-rate together.
Current fixed ops floorn/a~$25-$45/mo before DB overages and heavier LLM usageDepends on paid volume, not on launch countRailway Hobby/Pro and Vercel Pro are cheap enough that a small but real paid loop can cover them.
Break-even: Break-even is roughly 150 paid x402 calls per month, or about 24 marketplace copies at a $50 ticket with a 5% take. Launch fee alone does not break even unless it is repriced or launch volume is extreme.
Comparables
ComparableLaunchScale snapshotTime to scaleLesson
pump.fun2024-01-19DefiLlama-linked research showed roughly $523M annualized revenue by 2026 and $1B+ cumulative revenue within about two years.~24 monthsFrictionless launch + instant trading scales much faster than curated listings. But it also floods the market with low-quality supply.
Virtuals Protocol2024-03-14DefiLlama tracked roughly $69M annualized revenue, while the protocol whitepaper formalized agent launch fees, tax splits and graduation mechanics.~18-24 monthsCloser comp than classic launchpads: monetize the agent economy, not only token issuance. Fee splits and post-launch loops matter.
friend.tech2023-08-10Reached an early annualized revenue pace near $180M in 2023, then DefiLlama showed annualized revenue collapsing to ~$0 by 2026.~1 month up, ~2 years downSpeculative social loops can monetize attention immediately, but they decay just as fast without durable repeat utility.
Revenue scenarios
Realistic
12-18 months
1,500 paid x402 calls/mo + 400 marketplace purchases/mo at $12 average GMV + 100 launches/mo at a $0.50 fee
MRR: ~$1.0kARR: ~$12.5k
Optimistic
24-36 months
15,000 paid x402 calls/mo + 4,000 marketplace purchases/mo at $12 average GMV + 1,000 launches/mo at a $0.50 fee
MRR: ~$10.4kARR: ~$125k
Outlier
36-60 months · 5%
200,000 paid x402 calls/mo + 50,000 marketplace purchases/mo at $15 average GMV + 10,000 launches/mo at a $0.50 fee
MRR: ~$142.5kARR: ~$1.7M
Hard lesson
Hard lesson from creator-token infrastructure

Roll eventually closed its app after years trying to make creator-token infrastructure stick. The durable lesson is brutal: issuance is not retention. If users are paying only to mint or speculate, the revenue curve snaps when attention cools. Avax Radar needs repeat operator utility: API calls, copied strategies, paid skills and actual trading loops.

Open source note
Phase 4

Reality check anti-vanity

These are the default answers to the most likely strategic asks around this codebase.

Deploy deeper on mainnet right now
Only if it increases signed wallet usage in the next quarter. Otherwise it does not move revenue enough.
Minimum viable path: Keep the current C-Chain rails, harden the remaining EVM-only auth edge, and prove 10+ recurring paid wallets before spending more on operations.
Buy an external security audit now
No. That is vanity before recurring custodial volume.
Minimum viable path: Keep tightening the smallest risky surfaces: signed writes, signed vault reads, no fabricated discovery fallback, then one focused review when real volume appears.
Rewrite everything around fully autonomous wallets
No. Revenue does not require backend-owned keys for user wallets.
Minimum viable path: Sell the existing headless Agent API and let external agents sign their own actions. That captures the differentiation without inventing custody risk.
Package an SDK before usage proves out
No. Today the product is already consumable over HTTP + MCP.
Minimum viable path: Stabilize the routes, document auth, publish one tiny TS client only after real external usage appears.
Rely on the launch fee as the core business
No. At current code-level pricing that is not the revenue engine.
Minimum viable path: Keep launch friction low and push monetization toward x402 calls, copied strategies and paid skills where the take-rate compounds.
Phase 5

This page is the pitch surface

The market narrative now lives in product code, not in an external slide deck. It sits below Jury Flow because it is the investor and operator reality layer.

What this page does
Turns the strategic audit into a persistent route at /market
Keeps PT + EN narrative in one typed source of truth
Separates real rails from mock claims without killing the pitch
Gives judges and operators a market lens next to the demo flow
Sources
Vercel pricingRailway pricingGroq Llama 3.3 70B pricingAvalanche transaction feespump.fun metricsVirtuals launch mechanicsVirtuals metricsfriend.tech metricsRoll closure note