Skillv0.1.15

ClawScan security

HyperClaw · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 2, 2026, 2:55 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's code, environment requirements, and runtime instructions are consistent with a Hyperliquid trading tool; it legitimately needs a wallet address and API private key and installs a local caching proxy and Python dependencies.
Guidance: This skill appears coherent for trading on Hyperliquid, but it requires your HL_SECRET_KEY (a private key) — only provide an API wallet/private key with limited funds and permissions. Review the included scripts before running: setup.sh will pip-install packages into a created venv, and server.py runs a local caching proxy (default port 18731) that listens for requests. If you don't want a background server, you can skip starting it and point HL_PROXY_URL to the upstream API. Prefer running the skill in an isolated environment (container or VM), keep your main wallet keys offline, and audit optional intelligence integrations (XAI_API_KEY) before enabling them. If you want further checking, paste the remainder of hyperliquid_tools.py and server.py (they were truncated) so I can inspect any remaining handlers or external calls.

Review Dimensions

Purpose & Capability: okName/description (Hyperliquid trading, market scanning, intelligence) align with the requested secrets (HL_ACCOUNT_ADDRESS, HL_SECRET_KEY) and with the provided CLI and SDK-based implementation. The skill needs a private key to sign trades — appropriate for a trading agent.
Instruction Scope: noteSKILL.md instructs the user/agent to run setup.sh, create a venv, install packages, run a local caching proxy (server.py) and then invoke hyperliquid_tools.py commands. The runtime will read .env and a handful of environment variables. The instructions and code do not appear to request unrelated system files or arbitrary external endpoints beyond Hyperliquid and (optionally) an external intelligence API. The skill does use a local proxy and an optional Grok/XAI integration (XAI_API_KEY) for intelligence features — those are separate concerns and are documented as optional.
Install Mechanism: noteThere is no automated remote installer in the registry metadata, but the included setup.sh creates a Python venv and runs pip install -r requirements.txt. Dependencies come from PyPI (hyperliquid-python-sdk, eth-account, httpx, fastapi, uvicorn, etc.). This is standard but does execute package installs on the host; there are no downloads from ad-hoc URLs or extracts from personal servers in the manifest.
Credentials: noteDeclared required env vars (HL_ACCOUNT_ADDRESS, HL_SECRET_KEY) match the trading purpose. HL_SECRET_KEY is a sensitive private key and is legitimately required to sign orders; the README and SKILL.md explicitly recommend using a separate API wallet. The code also reads additional environment variables (HL_TESTNET, HL_PROXY_URL, HL_ENV_FILE, HL_PROXY_PORT, HL_UPSTREAM_URL, HL_CACHE_WARMUP) that are not listed in requires.env; some are optional and documented but HL_ENV_FILE and HL_PROXY_URL are not prominently documented in SKILL.md. Optional intelligence features reference XAI_API_KEY but it is optional.
Persistence & Privilege: okalways:false (no forced inclusion). The skill does not request to be permanently injected or to modify other skills. It does instruct running a background caching proxy (server.py) which will open a local HTTP endpoint and run while the user leaves it running — this is normal for the described caching behavior but is an extra persistent process you should be aware of.