Quant Trading CN
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This is mostly a quant-trading guidance skill, but it references missing or external executable trading tools and live broker automation that can affect real money.
Do not run the referenced scripts or upstream repository unless you review and trust that code. Start only in paper/backtest mode, avoid supplying live broker credentials until you have verified the implementation, and require explicit safeguards and capital limits before any real trading.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
66/66 vendors flagged this skill as clean.
Risk analysis
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.
The commands may simply fail, or a user/agent may end up running unrelated or unreviewed local scripts while working with financial-trading workflows.
The submitted manifest contains no scripts directory or code files, yet the skill instructs users or agents to run relative helper scripts. The executable behavior is therefore missing from the reviewed package and ambiguous.
./scripts/wizard.sh ... ./scripts/universe-fetch.sh --indices nifty50,nifty100,midcap150 ... ./scripts/check-code.sh ./my_trading_bot.py
Include the referenced scripts in the package for review, anchor commands to the skill install directory, or remove the executable instructions from this instruction-only package.
Following the docs could execute code that was not part of this review, including code capable of generating or running trading tools.
Embedded reference documentation directs users to clone and execute upstream code outside the submitted package, with no pinned commit or reviewed install specification in this artifact set.
git clone https://github.com/javajack/skill-algotrader.git algotrader cd algotrader ./start.sh wizard
Only run reviewed, pinned source code; the publisher should either vendor the code in the package or provide a pinned, auditable install spec.
A generated or copied bot could place real trades, close positions, or lose money without a per-trade human confirmation.
The guidance includes patterns for automatically placing live broker market orders. That is aligned with algorithmic trading, but it is high-impact financial account mutation and the artifacts do not define a clear live-mode approval or sandbox boundary.
if sl_modification_failures >= 3:
# Emergency: Close position with market order
kite.place_order(
order_type=kite.ORDER_TYPE_MARKET,Use paper trading by default, require explicit opt-in before live trading, set hard capital/risk limits, and require clear user confirmation before enabling broker order placement.
If these broker credentials are exposed or over-scoped, another party could access the trading account or place orders.
Live Zerodha trading credentials are purpose-aligned, and no leakage is shown, but the registry metadata declares no primary credential or required environment variables.
KITE_API_KEY=your_api_key KITE_API_SECRET=your_api_secret KITE_ACCESS_TOKEN=your_access_token
Keep credentials out of shared files and chats, use the least-privileged broker configuration available, rotate tokens regularly, and ensure .env files are ignored by version control.
Incorrect persisted state could cause duplicate trades, unmanaged positions, or wrong stop-loss handling.
The referenced bot design persists order and position state across sessions. This is normal for trading reconciliation, but stale or tampered state can influence later trading decisions.
"positions.json" structure ... "sl_order_id": "123456" ... "bot_managed": true
Store bot state securely, reconcile it against the broker before every live session, and review or reset state before switching from testing to live trading.