benchmark-praxis

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

• Third-party content exposure detected (high risk: 0.85). Yes—during required “Research” and “Feature delta/sweep” steps the workflow performs WebSearch/WebFetch and crawls docs pages/markdown from user-supplied URLs/repos, and those fetched outsider-authored page contents are then ingested into the agent’s LLM context for summarization and feature extraction (indirect prompt-injection risk).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly performs runtime WebFetch/GitHub API calls and raw GitHub fetches (e.g., https://raw.githubusercontent.com/{owner}/{repo}/{branch}/{path}, https://docs.claude.com/en/docs/claude-code/, gh api repos/{owner}/{repo}/git/trees/{branch}?recursive=1) and then injects that fetched documentation/markdown into per-fetch prompts to extract features, meaning remote content is fetched at runtime and directly controls the agent's prompt/context.