signals

Fail

Audited by Snyk on Apr 20, 2026

Risk Level: HIGH
Full Analysis

HIGH W007: Insecure credential handling detected in skill instructions.

  • Insecure credential handling detected (high risk: 1.00). The prompt includes explicit instructions and examples that place API keys directly into config files and curl headers (e.g., X-API-Key: bk_YOUR_KEY, config.json with apiKey), meaning an LLM generating those commands or code would need to output secret values verbatim — a high exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 1.00). The skill's required workflow (e.g., HEARTBEAT.md and SKILL.md) explicitly instructs the agent to poll public endpoints like https://bankrsignals.com/api/feed and https://bankrsignals.com/api/leaderboard (and accept webhook notifications) to consume user-published signals and reasoning, and those untrusted, user-generated signals are then used to drive copy-trading and other agent actions, so third-party content can materially influence behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).


MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

  • Direct money access detected (high risk: 1.00). The skill is explicitly built for crypto trading on the Base blockchain and includes wallet provisioning and a signing API (Bankr Agent API). It documents obtaining an API key (bk_...), provisioning EVM and Solana wallets, and calling a specific sign endpoint (https://api.bankr.bot/agent/sign) to produce EIP-191 signatures and signer addresses. The docs also reference "Sign & Submit API", "Leverage Trading", executing trades on Base, and workflows for publishing/closing trades with transaction hashes and collateral amounts. These are specific crypto/blockchain wallet and signing capabilities (i.e., crypto financial execution tools), not generic API or browser automation.

Issues (4)

W007
HIGH

Insecure credential handling detected in skill instructions.

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W012
MEDIUM

Unverifiable external dependency detected (runtime URL that controls agent).

W009
MEDIUM

Direct money access capability detected (payment gateways, crypto, banking).

Audit Metadata
Risk Level
HIGH
Analyzed
Apr 20, 2026, 06:19 AM
Issues
4