heurist-mesh

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt asks the user to provide an API key and includes examples (curl headers, config entries) that place that key directly into commands/configs, meaning an agent would need to output the secret verbatim to produce runnable requests—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 explicitly ingests and interprets public third-party content—see the Available Tools and Default Agents (e.g., exa_scrape_url "Scrape and summarize webpage content", twitter_search/user_timeline/tweet_detail, get_trending_tokens aggregating public sources, and Ask Heurist job results) which the agent uses to produce summaries and market intelligence, so untrusted web/social content can materially influence its actions.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill calls Heurist runtime APIs (e.g., https://mesh.heurist.xyz/mcp/ and https://ask-backend.heurist.xyz) to create/poll jobs and ingest returned result_text/agent outputs at runtime, meaning external content directly determines the agent's responses.