01
AI provider abstraction
Connect to OpenAI, Anthropic, a Codex subscription, or a local Ollama endpoint. AI inference billed by the operator to the operator's own provider account. No keys, prompts, or responses pass through localfile.studio.
In development · Signup list open
El Dorado Platform
A self-hosted markets terminal for crypto and prediction-market research.
Native desktop client. Bring your own AI provider. On-chain data, whale-flow analysis, contract risk screening, and Telegram-based remote control. Built in Rust for low latency and reliability.
Native desktop · BYO AI · Crypto + prediction markets · Telegram-ready
Why self-host
Most crypto trading software is a hosted dashboard with the operator's API keys on file. That model is fine when convenience is the only requirement. It is a worse fit when query privacy, vendor risk, AI cost control, or audit-trail integrity matter.
El Dorado runs as a native binary on the operator's machine. AI calls go through the operator's own provider account. On-chain queries originate from the operator's network. localfile.studio ships the binary; it is not in the loop on individual queries.
Capabilities
In development. The list below is the v1 scope.
02
Specialized research agents
Research, alerting, contract-risk, and execution-prep agents. Each agent has a defined scope, prompt template, and tool set. Multi-agent fan-in produces one structured response with citations.
03
Whale-flow analysis
Filter by chain, dollar threshold, time window, and counterparty. Plain-English queries supported. Drill into the on-chain transactions behind every total.
04
Contract risk screening
Paste a contract address. Returns liquidity depth, holder distribution, mint authority status, top-10 concentration, recent transfer patterns, and a structured risk score with the reasoning shown.
05
Telegram remote control
The terminal runs as a background service. A bring-your-own Telegram bot token provides two-way command and structured alerting. Mobile control without exposing keys.
06
Compiled native runtime
Rust binary. No garbage-collector pauses during agent fan-in. Memory-safe by default. Designed to run continuously for weeks without restart.
Common workflows
Three operating modes the terminal is built for.
New-token due diligence
Paste a contract; agents return a screening report (liquidity, holders, recent flow, risk markers) with cited sources.
Example query
“Research 0xA0b8…6eB48 for liquidity, holder concentration, and risk.”
Output
Structured report: LP depth across pools, top-10 holder share, mint authority status, recent transfer pattern flags, and a risk score with reasoning.
Whale-flow monitoring
Standing query, polled on a schedule, alert via Telegram when a configurable threshold is crossed.
Example query
“Alert if a wallet >$5M opens a new position on a token in the watchlist.”
Output
Telegram message with wallet address, token, position size, time, and a link to the underlying transactions.
Prediction-market event tracking
Watch Polymarket and Kalshi markets for probability changes, resolution windows, and liquidity shifts. Cross-reference on-chain context where relevant.
Example query
“Alert if Polymarket market X moves >5pp in 4h or liquidity exceeds $1M.”
Output
Probability series, current depth, time-to-resolution, and any on-chain settlement context the agents find relevant.
How it compares
vs. cloud terminals (CryptOn, HyperData, TradingView)
Cloud terminals optimize for setup time and collaboration. El Dorado optimizes for query privacy and provider independence. The operator owns the data path.
vs. subscription bots (3Commas, Cryptohopper)
Subscription bots accept exchange API keys and execute on behalf of the operator. El Dorado does not hold custody and does not sign trades. Execution stays in the operator's existing wallet or exchange.
vs. standalone Telegram bots
El Dorado includes Telegram as a remote-control surface, not the primary one. The desktop terminal is the workspace; Telegram is the mobile control surface.
vs. ad-hoc research (screenshots, threads)
Structured agent output with citations replaces ad-hoc research across chat threads and screenshots. The reasoning trail is preserved for audit.
Architecture
Three decisions worth naming.
Rust runtime
Compiled binary. No GC pauses. Memory-safe. The terminal is designed to run continuously for weeks under fan-in load from parallel agents and multi-provider RPC queries.
Multi-agent design
Each agent has a defined scope and tool set: research, alerting, contract-risk, execution-prep. A query is decomposed across the relevant agents, results fan back in, and the response is one structured object with citations.
Self-hosted by default
Binary on the operator's machine. SQLite for state. HTTP calls only to the operator's chosen AI provider and configured RPC endpoints. No phone-home. Telemetry opt-in, off by default.
Mobile control via Telegram
The terminal runs as a background service on the operator's machine. A bring-your-own Telegram bot token provides two interfaces: structured alerts (e.g., “watchlist token crossed 24-hour high”) and conversational commands (e.g., “research contract 0x... for liquidity and risk”). Alerts pause on schedule or on demand.
Telegram is the transport. The terminal does the work. No third-party server in between.
Self-hosting and data handling
- No custody. No transaction signing. Read-only on chain.
- No upload of watchlists, queries, or AI prompts to localfile.studio.
- Local SQLite for state; export anytime.
- Optional telemetry, off by default, fully visible in settings.
- If localfile.studio becomes unreachable, the terminal continues operating with configured providers.
Interface tour
Screenshots arrive with launch. Captions describe each view.
Screenshot 1 pending
Screenshot 2 pending
Screenshot 3 pending
Screenshot 4 pending
Screenshot 5 pending
Pricing announced at launch
El Dorado is in active development. Pricing is being calibrated against AI provider costs and data costs. The signup list gets the launch-day price first.
Frequently asked questions
When does El Dorado ship?
+
In development. The signup list receives one email at launch.
Why Rust?
+
Compiled binary, no GC pauses, memory-safe. The terminal is designed to run continuously for weeks under fan-in load.
How does bring-your-own AI work?
+
El Dorado does not include AI compute. The operator provides an API key (OpenAI, Anthropic, Codex subscription, or a local Ollama endpoint). The terminal calls the provider directly. No keys, prompts, or responses pass through localfile.studio.
Which chains and exchanges are supported on launch?
+
Coverage list published before pricing goes live. The architecture is chain-agnostic; the launch list reflects what is integrated and tested at ship time.
Does it hold keys or sign transactions?
+
No. El Dorado reads on-chain data, calls the operator's AI provider, and sends Telegram messages. It does not hold custody and does not sign trades. Execution stays in the operator's existing wallet or exchange.
How does Telegram integration work?
+
The operator provides a Telegram bot token. The terminal pairs with it for structured alerts and conversational commands. The bot process runs on the operator's machine; Telegram is the transport.
Where is data stored?
+
Local SQLite for watchlists, alert rules, agent conversation history, and cached on-chain queries. Export anytime.
Is this an investment advisor?
+
No. The agents return structured information; trading decisions remain the operator's.
Does El Dorado support prediction markets like Polymarket and Kalshi?
+
Yes. The agent and alerting system is venue-agnostic. Track probability changes, resolution windows, and liquidity on prediction-market venues alongside crypto tokens. Launch-day coverage list published before pricing.
Can the terminal cross-reference prediction-market positions with on-chain data?
+
Yes. For Polymarket markets with on-chain settlement, the agents can pull venue data and the underlying on-chain context into the same query.
How much will it cost?
+
Pricing announced at launch. The signup list gets the launch-day price first.
Will there be a free trial?
+
Launch model is one-time pricing with a refund window. Final terms with the launch announcement.
How is this different from CryptOn or HyperData?
+
Cloud terminals operate the data path. El Dorado operates locally. The AI calls bill against the operator's account. If the localfile.studio site becomes unreachable, the terminal continues running with configured providers.
Specifications
- Status
- In development. Signup list open.
- Price at launch
- TBD
- Language
- Rust
- Distribution
- Native binary (installer TBD)
- Platforms
- macOS, Windows, Linux
- Local storage
- SQLite
- AI providers
- OpenAI, Anthropic, Codex, local Ollama (BYO key)
- Markets supported
- Crypto tokens (multi-chain), prediction markets (Polymarket, Kalshi at minimum). Final launch list TBD.
- RPC providers
- BYO endpoints; one or many
- Telegram
- BYO bot token; one-way alerts + two-way chat
- Custody
- None. El Dorado never holds keys.
- Network calls
- On-chain RPCs and the operator's AI provider only. No phone-home.
- Update cadence
- Lifetime free version updates for V1 buyers (provisional)