Skillv1.0.0

ClawScan security

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

Scanner verdict

SuspiciousApr 18, 2026, 1:00 PM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill claims to fetch live market data from Yahoo Finance/CoinGecko/news but the code only calls an AISA LLM with a prompt (no direct web/data integrations) and also reads undeclared environment variables — this mismatch warrants caution.
Guidance: This skill will send whatever you run to the AISA API using the AISA_API_KEY you provide — it does not itself fetch Yahoo Finance, CoinGecko, or news via direct API calls, it asks the remote model to produce a live-scan report. Before installing: (1) Confirm you trust the AISA provider (https://api.aisa.one is the default base URL) and understand where your key will be sent. (2) Be aware the output may be hallucinated if the AISA model lacks real-time connectors; test with non-sensitive queries. (3) Note the script reads optional env vars AISA_BASE_URL and AISA_MODEL (undeclared in metadata) — review or sandbox the skill, and use a low-privilege or rotateable API key. (4) If you need guaranteed live data, prefer a skill that directly integrates with Yahoo/Coingecko APIs or other verifiable data sources.

Review Dimensions

Purpose & Capability: concernThe description promises 'live' scans of Yahoo Finance, CoinGecko, and news. The bundled script does not directly call those services; it only sends a prompt to an AISA model and asks the model to 'Use your built-in tools' to fetch live data. That is a capability mismatch unless the AISA model/system actually has connectors; the skill does not declare or implement any web-scraping or API calls to those sources.
Instruction Scope: concernSKILL.md instructs running the Python script which constructs an LLM prompt that requests live data. The runtime instructions do not read local sensitive files, but they delegate data collection to the remote LLM rather than performing direct, auditable fetches. This gives the model latitude to produce hallucinated 'live' data if it lacks real-time connectors.
Install Mechanism: okNo install spec; the skill is instruction + a small Python script. Nothing is downloaded or installed by the skill bundle itself.
Credentials: concernDeclared required env var is AISA_API_KEY (appropriate for calling the AISA service). However, the script also reads AISA_BASE_URL and AISA_MODEL (not declared in requires.env or documented in SKILL.md). The script will send the API key to the configured base URL (default https://api.aisa.one/v1). Undeclared optional env vars and sending an API key to a remote service should be noted.
Persistence & Privilege: okalways is false; the skill does not request persistent/always-on privileges, nor does it modify other skills or agent-wide configs.