jp-executor

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

• Third-party content exposure detected (high risk: 0.85). Yes—during required execution the skill re-fetches Juspay documentation at runtime via docs-mcp (doc_fetch_tool(url)), which ingests markdown text from outsider-authored public web pages into the agent’s LLM context for grounding.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill repeatedly requires runtime fetching of Juspay documentation via the docs-mcp doc_fetch_tool (allowed domains include *.juspay.io, *.juspay.in, dth95m2xtyv8v.cloudfront.net), and those fetched pages are a mandatory, runtime-controlled source that the agent uses to determine exact method/field names and implementation behavior.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly and specifically for integrating a payment gateway (Juspay) into a codebase: it implements a Juspay payment integration, provisions credentials/API keys, installs SDKs, writes core payment integration and order/payment schema, and implements webhook handlers and dashboard (portal) configuration. These are direct payment gateway integration actions (credentialed API/SDK usage and webhook/transaction handling), which match the "Payment Gateways" category of Direct Financial Execution.