YieldVault Agent

Autonomous yield farming agent for BNB Chain with deterministic execution, smart contract integration, and automated decision-making.

This skill contains the expected components for an autonomous on‑chain yield‑farming agent (decision engine, blockchain reader, tx executor, scheduler, Telegram alerts) and will need credentials and network endpoints to operate — but the registry metadata does not declare them. Before installing or running: - Do NOT run scheduler/tx-executor on mainnet or with real funds until you audit the code. Run everything offline or on a throwaway testnet account first. - Inspect package.json and package-lock.json to see what NPM packages will be installed. Audit dependencies and lockfile for unexpected packages or remote install scripts. - Grep the repository for process.env usage and for strings like PRIVATE_KEY, WALLET, TELEGRAM_TOKEN, BOT_TOKEN, RPC_URL, KEEPER_ADDRESS. Confirm exactly which env vars the code reads and how they are used. - Open tx-executor.js and scheduler.js and confirm where private key material is loaded and used. Prefer hardware wallet or KMS integration rather than a plaintext private key in env or config. If the code expects an env var for a raw private key, treat that as high risk. - Inspect notifications.js to confirm it only sends alerts to the configured Telegram API and does not exfiltrate data to unknown endpoints. Search for any hardcoded URLs, IPs, or base64 blobs in all files. - Verify contract addresses and ABI files are from trusted sources; artifacts can embed sensitive data or keys in rare cases. - Check INTEGRATION_GUIDE and config files to see whether any example configs include private keys; never use those directly. - If you plan to run on mainnet, require: hardware wallet/multi‑sig signing, Chainlink oracles as recommended, a formal smart contract audit, emergency pause mechanism, and minimal initial capital. Given the clear metadata/code mismatch (no declared required env vars vs code that needs secrets) and the pre-scan base64 indicator, proceed cautiously — treat this skill as 'suspicious' until you confirm how it handles secrets and network endpoints.

Type: OpenClaw Skill Name: yieldvault-agent Version: 1.0.0 The skill is classified as suspicious due to the high-risk vulnerability of directly using a private key from an environment variable (`WALLET_PRIVATE_KEY`) for signing blockchain transactions in `tx-executor.js` and configured in `config.scheduler.json`. While the documentation (e.g., `FINAL_CHECKLIST.md`, `RESPUESTAS_PREGUNTAS.md`, `SKILL_COMPLETION_REPORT.md`) explicitly acknowledges this as a critical security risk for mainnet and recommends hardware wallet integration, the current implementation presents a significant vulnerability if deployed without these mitigations. There is no evidence of intentional malicious behavior such as data exfiltration or unauthorized command execution; the code's functionality aligns with its stated purpose of autonomous yield farming.