retro

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly instructs the agent to "show" the whole observations file "unabridged" (and defaults to the whole file), which forces the LLM to output any secret values present in that file verbatim.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill explicitly directs the agent to edit and commit files across the user's home repos and dotfiles (e.g., "Make the change in this sitting" and "commit in each repo it touches"), which causes the agent to modify the machine's state.