sprite

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly instructs fetching tokens (from Keychain or user input) and interpolating them into commands (e.g., tmux send-keys, export CLAUDE_CODE_OAUTH_TOKEN, and examples with literal sk-... tokens), which requires the agent to include secret values verbatim in its output.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly captures and has the agent read InnerClaude's terminal output (tmux pipe-pane -> /tmp/claude-output.txt) and instructs OuterClaude to parse that output and make decisions (see "Quick Start" and "Working Loop"), and its bootstrap steps also pull public content (git clone, curl installs) from GitHub/URLs in references/setup.md, so untrusted third‑party content is fetched and directly influences subsequent actions.