Back to skill
Skillv1.0.0
ClawScan security
MegaETH AI Developer · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignFeb 11, 2026, 9:02 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill is internally consistent with its stated MegaETH developer purpose and contains no unexpected credential or install demands, but several operational recommendations in the instructions are risky and deserve developer caution.
- Guidance
- This skill appears to be what it claims (MegaETH developer playbook), but you should: 1) Verify the skill's origin before trusting it — registry metadata lacks a homepage while README points to external repos; clone/inspect the repo yourself. 2) Review the docs' risky operational advice (skipping local simulation, hardcoding gas limits, pre-signing/pipelining transactions) and do NOT apply those recommendations to high-value wallets or production without thorough testing. 3) Always test on the stated testnet (chain ID 6343 / carrot.megaeth.com) and run mega-evme locally to replay and simulate transactions when possible. 4) Use trusted RPC endpoints (avoid random public RPCs), protect private keys (never allow the agent to access them), and avoid pre-signing with hot keys or shipping pre-signed transactions without manual review. 5) Confirm external URLs (docs.megaeth.com, mainnet.megaeth.com, Kyber aggregator endpoints, and GitHub repos) are legitimate and up-to-date before following install/build commands.
Review Dimensions
- Purpose & Capability
- okName/description and the files (wallet operations, RPC methods, gas model, debugging, etc.) align: everything requested and documented is relevant to building on MegaETH. Note: the skill's source/homepage is unknown in registry metadata while README references GitHub/Clawdhub install commands — verify origin before trusting.
- Instruction Scope
- noteSKILL.md and companion docs focus tightly on MegaETH tasks (RPC, WebSocket, gas, storage, debugging). However, several recommendations are operationally risky: advising to 'skip local simulation', 'hardcode gas limits', 'pre-sign transactions' and 'nonce pipelining' can lead to lost funds or unsafe deployments if applied blindly. The SKILL also expects the agent to produce diffs and commands (reasonable) but does not request unrelated files/credentials.
- Install Mechanism
- okInstruction-only skill with no install spec or code to download; lowest install risk. README mentions git/ npx/ clawdhub install flows but the registry package itself has no install actions.
- Credentials
- okNo required env vars, binaries, or credentials are declared or requested. References to RPC endpoints and third-party services (MegaETH RPC, Kyber aggregator, GitHub repos) are expected for this domain and proportional to the stated functionality.
- Persistence & Privilege
- okSkill does not request persistent or elevated privileges (always:false, no config paths). It is user-invocable and can be used autonomously per platform defaults; nothing here demands forced or system-wide persistence.