Quant Trading CN
WarnAudited by ClawScan on May 10, 2026.
Overview
This looks like trading-bot guidance rather than malware, but it deserves review because it can lead an agent to use broker credentials and place live market orders while the advertised helper scripts are not present for review.
Use this skill only for education, code review, or backtesting unless you intentionally want live broker automation. Do not provide Kite/Zerodha credentials to generated code until you have reviewed it, tested in paper mode, set strict limits, and enabled manual approval for live orders. Be aware that the advertised scripts are not included in the reviewed package.
Findings (5)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
An agent-generated or externally fetched bot could place, cancel, or close real trades, potentially causing financial loss.
The skill's reference material includes patterns for placing live market orders through a broker API. That is purpose-aligned for a trading bot, but it can mutate a financial account and the provided artifacts do not show explicit per-order approval, paper-trading defaults, or hard loss limits.
Emergency: Close position with market order ... kite.place_order(order_type=kite.ORDER_TYPE_MARKET, quantity=position['quantity'], ...)
Use paper trading by default, inspect any generated code, require explicit human confirmation for every live order, and set strict broker-side capital and loss limits.
Anyone or any code with these credentials may be able to access the brokerage account and place trades within the token's permissions.
The live-trading setup asks for broker API credentials and an access token. The registry metadata says there is no primary credential and no required environment variables, so this high-impact account access is under-declared.
KITE_API_KEY=your_api_key KITE_API_SECRET=your_api_secret KITE_ACCESS_TOKEN=your_access_token
Do not provide live broker tokens unless you intend real trading; use least-privilege or paper-trading credentials, rotate tokens, and have the skill metadata declare the credential requirements.
The skill may fail as documented or encourage a user/agent to fetch and run unreviewed upstream code to obtain the missing trading automation.
The supplied package has no executable helpers, while SKILL.md instructs use of ./scripts/wizard.sh, ./scripts/universe-fetch.sh, and ./scripts/check-code.sh. The advertised automation is therefore missing from the reviewed artifacts.
No install spec — this is an instruction-only skill. ... No code files present — this is an instruction-only skill.
Include the referenced scripts in the reviewed package or remove those commands; if external code is required, pin the source and version and review it before execution.
Users may over-trust generated strategies and deploy them with real capital without adequate validation.
The skill advertises a specific win-rate signal. SKILL.md also includes a trading-risk disclaimer, so this is not deceptive by itself, but users should not treat the claim as a guarantee.
📈 65% 胜率信号
Treat performance claims as educational examples only; independently backtest, paper trade, and validate risk assumptions before live use.
Local files may reveal sensitive trading activity or cause a bot to make incorrect decisions if they become stale or are modified.
The guidance recommends persistent local trading state containing position and order details. This is expected for bot reconciliation, but stale or tampered state can influence future trading decisions.
"positions.json" ... "quantity": 100, "entry_price": 1500.00, "sl_order_id": "123456", "bot_managed": true
Protect local state files, keep them out of source control, reconcile with the broker as the source of truth, and review state before live trading.