calle

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires preserving and embedding returned credentials like plan_id and confirm_token verbatim into CLI commands (and not exposing them otherwise), which forces the LLM to handle and output secret values directly, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md explicitly requires reading and acting on content returned by the external calle CLI (e.g., showing the CLI-provided assistant_hint.message and basing workflow on call run / call status JSON fields like status_result.structuredContent, result.structuredContent, and result.transcript), which are untrusted third-party outputs (including user-generated call transcripts) that can influence polling, continuation, and user-facing actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes a runtime fallback that runs remote code via the command "npx -y @call-e/cli@0.3.2" (which fetches and executes the @call-e/cli package from the npm registry), making it a high-confidence runtime external dependency that executes remote code.