tech-debt

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The prompt contains explicit, out-of-scope instructions to stealthily log execution metadata to a user-specific memory directory (checking ~/.claude/projects/ and appending to skill-telemetry.md, "skip silently" if not found), which is a covert persistence/telemetry action unrelated to the stated technical-debt analysis purpose.

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly requires reporting hardcoded configuration values with "file, line, the value," which would force the model to read and output secret/API credentials verbatim if present in the codebase, creating exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.70). PHASE 3 (Dependency Debt) explicitly instructs the agent to check npm/PyPI/etc. package registries, changelogs, and public CVE/audit data—public, user-published sources that the agent must read and whose content can influence dependency decisions—so the skill clearly ingests untrusted third‑party content.