huawei-cloud-skill-tester

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill repeatedly requires prompting the user for Huawei Cloud AK/SK and instructs setting them as environment variables/using them to run real CLI/SDK steps, which forces the agent to accept and potentially emit those secret values (e.g., in exported env vars, commands, or logs), creating an exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The pipeline intentionally executes skill-supplied Python/snippets (RCE) while collecting/exporting cloud AK/SK into the environment and even auto-confirms write operations / cleans up artifacts — a design that enables credential exposure, remote code execution, and covert resource modification or data exfiltration.

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

• Third-party content exposure detected (high risk: 0.75). This workflow ingests outsider-authored free text from other skills’ local SKILL.md files (and their embedded code/markdown blocks) at runtime via Tier 1 Phase 1 parsing (scripts/tier1/phase-1-skill-analysis.sh reads and extracts content from ${skill_dir}/SKILL.md), then propagates it into later JSON that may be consumed by the reporting logic.