senpi-entrypoint

Warn

Audited by Snyk on Mar 11, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.90). The skill mandates a Pre-Response Invocation Check and includes a background checker (scripts/check-skill-updates.py) that fetches GitHub repo listings via the GitHub API and raw SKILL.md files from https://raw.githubusercontent.com, and the agent parses those untrusted, public SKILL.md contents to decide and surface "new" or "updated" skills and to drive install/recommendation actions—so third-party content from GitHub is fetched, interpreted, and can change agent behavior.

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

  • Potentially malicious external URL detected (high risk: 0.90). The skill invokes runtime installs and live checks that fetch and parse content from the Senpi GitHub repo (e.g. https://github.com/Senpi-ai/senpi-skills and its raw files on raw.githubusercontent.com) — these npx install commands and the check-skill-updates.py fetch SKILL.md content at runtime which directly controls agent instructions/skill behavior, so this is a high-confidence external control dependency.

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

  • Direct money access detected (high risk: 1.00). The skill is explicitly for onboarding agents into a trading platform (Senpi on Hyperliquid). It walks users through account creation, API key generation, MCP server configuration, and "first-trade guidance," and equips the agent with a "trading toolkit" and installable trading skills (e.g., strategy skills). These steps and artifacts are specifically designed to enable market trading (crypto/asset execution) rather than being generic tooling. Although the file doesn't list a single "place_order" call, the primary and explicit purpose is to set up and enable trading (crypto/market execution) via API keys and platform configuration, which gives the agent direct financial execution capability once combined with the trading skills it installs.

Issues (3)

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
MEDIUM
Analyzed
Mar 11, 2026, 07:34 PM
Issues
3