research

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill embed literal API keys, bearer tokens, and other secrets directly in example curl commands and requests, which requires the LLM to handle or reproduce secret values verbatim and thus creates a high exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This document contains multiple hard-coded API keys/bearer tokens and a centralized secrets path, representing credential leakage that enables unauthorized access and targeted data collection/abuse.

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

• Third-party content exposure detected (high risk: 0.85). The required workflow in SKILL.md instructs runtime web/news/social searches (e.g., Brave/Exa/NewsAPI/GNews/X/Twitter) whose fetched page/article/tweet text would be ingested into the agent’s LLM context, creating an indirect prompt-injection path from outsider-authored free text.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). Multiple literal, high-entropy API keys/tokens appear in the documentation and are not placeholders or obvious examples. I inspected each potential match and confirmed they are real-looking credentials (not "YOUR_*" placeholders, not trivial setup passwords, not truncated/redacted). Specifically:
• SKILL.md:10 (X-Subscription-Token) — high-entropy token form, not a placeholder.
• SKILL.md:16 (Exa x-api-key) — UUID-like API key, literal value.
• SKILL.md:24 (Twitter bearer token) — long encoded bearer token, high entropy.
• SKILL.md:29 (Google API key) — begins with "AIza", matches Google API key pattern; matches the provided prefilter.
• SKILL.md:34 (NewsAPI apiKey) — hex-like, high-entropy API key.
• SKILL.md:39 (GNews token) — alphanumeric token, high-entropy.
• SKILL.md:56 (Apollo X-Api-Key) — nontrivial API key string.

None of these meet the "ignore" criteria (they are not placeholders like YOUR_API_KEY, not simple example passwords, not truncated/redacted markers). Therefore they should be flagged as secrets.