Utilia Agent Media and Solana Tools

Utilia normalizes audio, converts public PDFs to agent-ready Markdown, and gives Solana transaction bots live, account-localized transaction intelligence. Each tool accepts exact x402 USDC payments without accounts, subscriptions, or API keys.

Audio normalization

POST /v1/audio/normalize accepts one public HTTPS audio URL or base64 audio payload and returns a bounded 128 kbps MP3 as base64. The $0.01 response includes input and output digests, output duration, truncation status, and FFmpeg loudness measurements. The default target is -16 LUFS; callers can select -24 through -12 LUFS and cap output between 1 and 180 seconds.

{
  "url": "https://example.com/voice-note.wav",
  "targetLufs": -16,
  "maxSeconds": 180
}

PDF to Markdown in one command

export SOLANA_KEYPAIR_PATH=/absolute/path/to/agent-wallet.json
npx -y utilia-solana-agent pdf-to-markdown https://example.com/document.pdf

The $0.0025 response includes page-delimited Markdown, page count, metadata, and a source SHA-256 digest. Install the dedicated agent skill with:

npx skills add mohamedkuch/utilia-solana-agent \
  --skill utilia-pdf-to-markdown -g -y

The same skill is browsable on skills.sh and Smithery.

Live Solana priority fees

For the shortest wallet-managed flow, AgentCash can inspect the live quote, pay the x402 challenge, and retry the request:

npx -y agentcash@latest check https://api.utilia.ink/base/v1/fees/priority
npx -y agentcash@latest fetch https://api.utilia.ink/base/v1/fees/priority --yes

Run a JSONL fee watcher at five calls per hour, capped at 25 calls and $0.05:

export SOLANA_KEYPAIR_PATH=/absolute/path/to/low-balance-agent-wallet.json
npx -y utilia-solana-agent watch-fees --every 12m --max-calls 25

Product

Endpoint	Price	Result
GET /v1/transaction/:signature	$0.004	Transaction outcome, deltas, logs, compute use, and error classification
POST /v1/transaction/simulate	$0.008	Pre-broadcast simulation and failure explanation
GET /v1/fees/priority	$0.002	Recent priority-fee quantiles
POST /v1/token/risk	$0.006	Workflow-friendly mint safety check with a JSON { "mint": "..." } body
GET /v1/token/:mint	$0.006	Mint controls, Token-2022 features, and concentration signals
POST /v1/pdf/to-markdown	$0.0025	PDF text, page-delimited Markdown, metadata, and source digest
POST /v1/audio/normalize	$0.01	Bounded normalized MP3, loudness measurements, and content digests

All six products are also available as paid tools from the remote Streamable HTTP MCP server at https://api.utilia.ink/mcp. Its public server metadata is available at https://api.utilia.ink/server.json.

The server is published in the official MCP Registry as ink.utilia/solana-preflight. The public API contract is available at https://api.utilia.ink/openapi.json, and live availability is reported by https://api.utilia.ink/healthz.

The priority-fee, PDF-to-Markdown, token-safety, and transaction-simulation products are ownership-verified and immediately runnable through x402.jobs. The live fee feed is listed on both Solana and Base; the PDF converter is likewise listed on Solana and Base. The workflow-friendly token-safety resource is available on Solana and Base; both accept a Solana mint as JSON. Agents can also call the public audio-normalization resources on Solana or Base, the Solana Priority Fee Oracle workflow, or the Solana Token Safety Check workflow. The public Normalize Audio to MP3 workflow accepts a public audio URL and passes the normalized MP3 result through to the buyer. Agents can call its paid webhook directly at https://api.x402.jobs/@utilia/normalize-audio-to-mp3-dnx1du.

Each HTTP product also has a /base/v1/... alias for Base USDC buyers and Coinbase Agentic Market discovery. The canonical /v1/... routes and MCP tools continue to settle on Solana.

For agents that prefer a marketplace purchase flow, Utilia also has native PayanAgent offers, including audio normalization, and four healthy, instant-delivery services in the402 catalog: priority fees, Solana token safety, PDF-to-Markdown, and FFmpeg audio normalization.

Installable agent instructions are publicly available through ClawHub, including the dedicated audio-normalization skill, and through skills.sh and Smithery. Agent Tools Cloud indexes the direct MCP record and the Solana priority-fee route.

The PDF and audio transforms accept either a public HTTPS URL or a base64 payload, reject private-network fetches, and cap input at 8 MiB. Audio output is additionally capped at 5 MiB and 180 seconds; PDF input is capped at 100 pages and explicitly reports that scanned image-only documents require OCR. The public homepage, health check, and OpenAPI document are free. A paid request that ends in a 4xx or 5xx response is canceled before x402 settlement.

For a guarded audio conversion that verifies the returned digest and writes the MP3 without overwriting an existing file:

npx -y utilia-solana-agent@0.5.5 audio-normalize <public-https-audio-url> \
  --output normalized.mp3

Agent quick start

An x402-capable MCP agent connects to https://api.utilia.ink/mcp, lists the six tools, and pays automatically only when it calls one. A runnable client with a strict maximum-payment policy is included in examples/mcp-client.mjs.

Remote MCP configuration:

{
  "mcpServers": {
    "utilia": {
      "type": "streamable-http",
      "url": "https://api.utilia.ink/mcp"
    }
  }
}

SOLANA_PRIVATE_KEY='base58-wallet-private-key' \
  node examples/mcp-client.mjs solana_priority_fees '{"accounts":[]}'

Local development

Requirements: Node.js 22 and npm.

npm ci
HELIUS_API_KEY=your-key PAYMENTS_ENABLED=false npm run dev

Then open http://localhost:4021. Payments are deliberately disabled for local tests.

Run the complete verification suite:

npm test
npm run test:coverage
npm run build
npm run lint
npm run format:check
npm audit

Production model

• Solana RPC is supplied by Helius.
• Exact USDC payment verification and settlement initially uses the PayAI x402 facilitator. FACILITATOR_MODE controls the canonical Solana routes and MCP, while EVM_FACILITATOR_MODE independently lets the /base/... aliases use CDP. The two middleware stacks never advertise a network through the wrong facilitator.
• Bazaar discovery metadata is attached to every Solana and Base paid route.
• The remote MCP server is published through the official MCP Registry.
• Caddy terminates TLS and proxies only to a loopback-bound Docker port.
• The container runs as an unprivileged user with a read-only filesystem, no Linux capabilities, bounded memory/CPU/PIDs, and rotated logs.
• The server stores no wallet private keys. SVM_PAY_TO and EVM_PAY_TO are only public receiving addresses.

See Architecture and Operations.

Economics

Revenue is usage multiplied by endpoint price, less facilitator fees, infrastructure, RPC costs, taxes, refunds, and payment conversion costs. This software creates a sellable service; it does not guarantee demand or profit. The initial prices remain above PayAI's published post-free-tier per-settlement fee and should be revisited using real conversion, retention, and endpoint-cost data.

License

Source available for inspection. Copyright © 2026 Mohamed Kechaou. All rights reserved; no license is granted to copy, modify, or redistribute the server source.