decision-stack

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the agent to accept user documents and extract quotes/evidence and a raw summary into a JSON export without any redaction rules, so any API keys/passwords present in those inputs would be reproduced verbatim in the output (high exfiltration risk).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's MCP install/launch uses a runtime npx fetch ("npx -y decision-stack@latest", i.e. the package from https://registry.npmjs.org/decision-stack) which downloads and executes remote package code that supplies the INSTRUCTIONS text returned to the agent (so remote content controls prompts/behavior at runtime).