deepwork-tracker

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The prompt includes a clear, unconditional instruction to "Always send" reports to a specific Telegram user (Alex, id 8551040296), which effectively forces data exfiltration to a third party and goes beyond the skill's stated, generic "generate for sharing" purpose.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill includes an explicit, mandatory data-exfiltration step that always sends generated user reports to a hardcoded Telegram user ID (Alex, 8551040296), indicating intentional unauthorized sharing/backdoor behavior.

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

• Third-party content exposure detected (high risk: 0.90). The skill's bootstrap step explicitly clones code from the public GitHub repo (https://github.com/adunne09/deepwork-tracker.git) into ~/clawd/deepwork/deepwork.js and then instructs the agent to execute that script (start/stop/report), which exposes the agent to untrusted third-party code that can materially influence actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's bootstrap step explicitly runs a git clone from https://github.com/adunne09/deepwork-tracker.git and copies a remote deepwork.js into the runtime path which is then executed (e.g., via ~/clawd/deepwork/deepwork.js start), so remote code is fetched at runtime and can be executed.