Skillv1.0.0

ClawScan security

vibetrading-ai-trading-code-generator · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousFeb 13, 2026, 12:21 PM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The package’s stated purpose (generate Hyperliquid trading strategies) mostly matches its included files, but there are multiple inconsistencies, broken/unsafe implementation details, and missing declarations that make this skill risky to run without review or sandboxing.
Guidance: This skill is functionally aligned with its purpose but has several red flags and outright coding issues — do not run it with live exchange credentials or on an unrestricted machine yet. Specifically: - Inspect api_wrappers/hyperliquid_api.py carefully to verify where network requests go and whether secrets are used or exfiltrated. Ensure endpoints are legitimate. - Review code_validator.py: it creates and runs a temporary script that imports target modules. That import may execute arbitrary code with side effects (network calls, filesystem, order placement). Avoid running the validator on untrusted/generated modules until audited, or run it inside an isolated sandbox/container without network access. - Fix the obvious mismatches and syntax issues (examples: malformed return values like {"filepath"(filepath)}, missing imports of Path in validator, and truncated/ malformed blocks). Several files contain syntax/logic errors and will likely crash or behave unpredictably. - Do not provide production Hyperliquid API keys or account addresses to this code until you (or a trusted developer) have audited and corrected the code and run it in a testnet/sandbox environment. Prefer testnet API keys, or mock the API layer when validating/generating code. - Ask the publisher (or request an updated package) to: (1) declare required env vars in the registry metadata, (2) fix SKILL.md to reference the actual script names, (3) remove or clearly document any validation steps that import/execute user code, and (4) fix the syntax/logic errors before using in production. If you want, I can: point out specific lines with likely bugs, list the exact files that appear to execute imports/subprocesses, or draft an isolated test plan (container + dummy keys) you can use to safely evaluate the skill.

Review Dimensions

Purpose & Capability: noteThe skill claims to generate executable Hyperliquid strategies and includes templates, a Hyperliquid API wrapper, price fetcher, and generator/formatter/validator scripts — which is coherent with the description. However the SKILL.md and README reference scripts (e.g., scripts/strategy_generator.py) that are not present in the file manifest (the repo contains strategy_generator_with_prices.py and other similarly named files). Also templates and examples expect environment variables (HYPERLIQUID_API_KEY, HYPERLIQUID_ACCOUNT_ADDRESS) and API access even though the registry declares no required env vars; that mismatch should be resolved.
Instruction Scope: concernRuntime instructions tell the agent to run generator, validator, price fetcher, and other scripts. The code_validator creates and executes a temporary Python script which imports the target module (import {module_name}) to detect import/runtime errors — this will execute arbitrary module code during validation (side effects, network calls, order placement, etc.). That's expected for an aggressive import-based import-checker but is dangerous if untrusted code (templates or generated strategies) is present. Additionally, SKILL.md commands reference a missing script name (strategy_generator.py), giving the agent ambiguous instructions. Several files and validator behavior refer to writing and executing generated code in generated_strategies/ and running backtests — all of which can cause network connections, filesystem changes, or order placements if run with live credentials.
Install Mechanism: okNo install spec is provided (instruction-only), so the skill does not download or install external binaries automatically. That lowers installation risk. It does require python3 to be present as declared.
Credentials: concernThe code and templates expect Hyperliquid API credentials to be provided via environment variables (HYPERLIQUID_API_KEY, HYPERLIQUID_ACCOUNT_ADDRESS) and the SKILL.md's validation rules explicitly check for API keys loaded from env. Yet the registry lists no required env vars or primary credential. This omission is a coherence issue: the skill will prompt for/expect sensitive credentials at runtime but does not declare them. You should expect to supply API keys to actually run strategies; do not provide live keys until code is audited and executed in a safe environment.
Persistence & Privilege: okThe skill is not marked always:true and does not request unusual system privileges in the manifest. It will create directories (generated_strategies, logs, sessions, simulation_results) and write files; that file-write behavior is consistent with a code generator. Nothing in the manifest indicates modification of other skills or system-wide configuration.