lcx-doctor

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill instructs the agent to capture and cite command output and file contents (including $CODEX_HOME/config.toml and stderr) verbatim as evidence, which can contain API keys or secrets and there's no instruction to redact them.

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

• Third-party content exposure detected (low risk: 0.10). The required workflow only reads first-party provider code by cloning GitHub repos specified in the workflow into /tmp and then inspects local files/config on the host (e.g., ~/.codex/config.toml, installed plugin payload), without ingesting arbitrary outsider-authored free text without selecting specific repository content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly clones GitHub repositories at runtime (sync_latest_source is invoked for code-yeongyu/lazycodex and openai/codex), and those remote checkouts are read and relied on to drive the diagnostic comparisons, so external repo contents directly influence agent behavior.