jobs-interview

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). Both URLs are raw GitHub-hosted install scripts that the skill instructs to pipe directly into bash/PowerShell — executing remote scripts from an unverified repository is a high-risk vector for malware distribution.

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md, the workflow reads outsider-authored offer text from a cached offer using sjctl offer show <offerKey> --json (and then generates questions and saves them), so attacker-submitted free text can be consumed at runtime if it was inserted into that cache by a prior /jobs-search flow.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill includes installation commands that fetch and execute remote scripts at runtime via "curl -fsSL https://raw.githubusercontent.com/solid-company/solid-jobs-skills/v0.6.0/scripts/install-sjctl.sh | bash" and "irm https://raw.githubusercontent.com/solid-company/solid-jobs-skills/v0.6.0/scripts/install-sjctl.ps1 | iex", which download and run code required for the skill if sjctl is not present.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill tells the agent to install a remote CLI by piping a downloaded script into a shell/PowerShell (curl | bash, irm | iex) which modifies the host system and could run with elevated privileges, so it encourages changing machine state even though it doesn't explicitly request sudo.