Best Finance & Trading Skills for Claude Code

Solana DEX swap execution via Jupiter aggregator including quoting, transaction building, signing…

Best-in-class trade-execution skill: a 7-step Jupiter swap pipeline with hard confirmation gates ('NEVER proceed without explicit yes', steps 2-3 'never skip'), 8 non-negotiable safety requirements (default simulation, never store/log private keys, never 100% slippage, price-impact block thresholds, SOL reserve), and a dedicated safety-checklist reference with block/warn/override tiers. simulate_swap.py explicitly never signs or submits and requires no keys; get_quote.py degrades gracefully without HELIUS_API_KEY. 6.0 manifest: network destinations quote-api.jup.ag / price.jup.ag (official Jupiter), api.mainnet-beta.solana.com (official Solana RPC), mainnet.helius-rpc.com (optional, official) - all declared; WALLET_PRIVATE_KEY read from env only at signing time in the documented flow, never stored/logged; no persistence. Capabilities match declared scope; risk gating for funds-spending execution is exemplary.

Broad crypto market data from CoinGecko covering 13,000+ tokens. Global market stats, historical …

Exemplary tool-wrapper skill: clear When-to-Use and explicit When-Not (defers to Birdeye/DexScreener for real-time Solana data), quick-start snippets, free/pro auth handling, rate-limit backoff code, honest limitations section, demo-mode scripts. Only minor gap: no trigger-phrase variants in description.

Jito bundle submission for MEV protection on Solana — bundle building, tip strategies, block engi…

Trade-execution skill for Jito bundle submission on Solana, with strong risk gating for a funds-spending skill: prominent SAFETY WARNING, --demo default mode, MAX_TIP_LAMPORTS 0.01 SOL hard cap, pre-submission safety checklist, and live mode deliberately does not submit (no wallet key handling). Scripts runnable with argparse, help text, and mock responses. Minor: SKILL.md snippets have small bugs (undefined keypair in build_protected_swap_bundle, missing asyncio import in submit_with_retry). 6.0 manifest: network destinations limited to the four official *.block-engine.jito.wtf endpoints matching the declared purpose; env reads JITO_BLOCK_ENGINE/SOLANA_RPC_URL declared; no credential access, no file writes, no persistence. Capabilities match declared scope; tip accounts listed are the public Jito tip accounts.

Kalshi exchange mechanics — RSA-PSS auth, order schema, YES/NO orderbook convention, WebSocket, a…

Excellent tool-wrapper reference for the Kalshi exchange API. Strong: RSA-PSS signing spec with the three classic 401 failures, dollar-string order schema with per-field 400 rules, YES/NO both-bid-ladder convention, WS signing-path gotcha, T-1 discovery pipeline, dead-host warnings, verify-before-coding guard. Script is pure stdlib offline with worked example. Deductions: D1 lacks explicit when-not; D3 order-placement is documented without an explicit pre-trade confirmation gate (acceptable for a reference wrapper, but noted). 6.0 manifest: network destinations api.elections.kalshi.com / external-api-ws.kalshi.com / demo-api.kalshi.co (all official Kalshi), docs.kalshi.com, trading-api.readme.io (official legacy mirror), github.com/Kalshi - all declared exchange endpoints, no undeclared egress. Reads KALSHI_KEY_ID / KALSHI_PRIVATE_KEY_PATH from env with explicit 'never in code' guidance - legitimate, declared credential use. No injection, no destructive commands, no persistence.

Free DeFi analytics across all chains — TVL, token prices, DEX volumes, fees/revenue, stablecoins…

Clean tool-wrapper for DeFiLlama: base-URL/auth table per service, endpoint docs with response shapes, coin-identifier reference, free-vs-Pro endpoint matrix, and a when-to-use-vs-alternatives routing table. Scripts call only official *.llama.fi hosts (keyless). 6.0 manifest: network destinations api.llama.fi / coins.llama.fi / stablecoins.llama.fi / bridges.llama.fi only (all official DeFiLlama, matching declared purpose); no credentials, no file writes, no persistence; capabilities match declared scope. Read-only macro-analytics skill, no execution or advice risk.

Free, no-auth multi-chain DEX pair data — prices, volume, liquidity, transactions, and token prof…

Clean tool-wrapper for the free no-auth DexScreener API: endpoint surface, full pair-response schema with type gotchas (string prices, ms timestamps), common analysis patterns, explicit Limitations section, and a when-to-use-vs-alternatives routing table (Birdeye/Helius/SolanaTracker/Yellowstone). Scripts call only api.dexscreener.com with timeout and exception handling. 6.0 manifest: network destinations api.dexscreener.com only (official, matches declared purpose); no credentials, no file writes, no persistence; capabilities match declared scope. Read-only market-data skill, no execution, no advice risk.

Systematic exit rules, stop-loss methods, take-profit strategies, and trailing stop implementatio…

Systematic exit-rules skill: six exit categories (stop loss, take profit, trailing, time, signal, liquidity) with code, a layered exit-plan template with explicit priority hierarchy (hard stop never overridden), scaled-exit tranches, PumpFun-specific decay rules, and a common-mistakes table. Both scripts are offline (synthetic data / pure math) and carry 'informational only, not financial advice' disclaimers. 6.0 manifest: no network destinations, env vars used only for calculator inputs, no credential access, no file writes, no persistence; capabilities match declared scope. Compliance-positive: risk-first framing and no-advice disclaimers.

Feature construction from market data for ML trading models including price, volume, on-chain, an…

High-quality ML feature-engineering skill for trading models: 7 feature categories with formulas and lookbacks, stationarity (ADF) testing, no-lookahead rules stated as absolute constraints with correct/incorrect code contrast, 4-step feature-selection pipeline, label creation guidance, and two scripts (demo default with synthetic data; --live via Birdeye). Cross-skill integration map is well done. 6.0 manifest: network destination public-api.birdeye.so only, in explicitly opt-in --live mode, with BIRDEYE_API_KEY read from env and sent only to Birdeye itself as auth header - declared in both SKILL.md and script docs; no local file writes, no persistence. Capabilities match declared scope. No injection, no secrets hardcoded.

Impermanent loss calculation, modeling, and breakeven analysis for AMM liquidity provision across…

Solid quant skill: constant-product and CLMM IL formulas with derivations, concentration-factor approximation with explicit validity bounds, sigma^2/8 breakeven framework, Monte Carlo scenario modeler, and clear When-acceptable/When-to-avoid sections. Scripts are pure-stdlib/pure-numpy offline calculators with argparse and demo modes. Deductions: description is a single generic sentence without when/when-not (boundaries live only in SKILL.md body); a bit of conceptual background at the top. 6.0 manifest: no network, no env/credential reads, no file writes, no persistence; capabilities match declared scope. Analytics only - no trade execution, no advice disclaimers needed; risk framing ('when to avoid LPing') present.

AMM pool mechanics comparison across Raydium, Orca, and Meteora including fee structures, pool ty…

Comprehensive Solana AMM pool-mechanics reference: per-DEX program IDs and fee tiers (Raydium V4/CLMM, Orca Whirlpool, Meteora DLMM/Dynamic, PumpSwap), PumpFun graduation lifecycle, volume-efficiency interpretation bands, red-flag table, a 5-component health-score algorithm, and execution-pool ranking. Scripts fetch only DexScreener (keyless) with a --demo mode. No-advice disclaimer at the end of SKILL.md. 6.0 manifest: network destinations api.dexscreener.com / dexscreener.com links only (official, declared); no credentials, no file writes, no persistence; capabilities match declared scope. Analysis-only skill, no trade execution.