The Agent Skills Directory

[SAFE]: The skill implements a standard research and summarization workflow using authorized search and fetch tools. No malicious behavior or indicators of compromise were found.
[PROMPT_INJECTION]: There is a potential indirect prompt injection surface as the skill fetches untrusted web content via web_search and fetch_url. This is a known risk for research tools but is considered safe here because the skill lacks privileged capabilities and is restricted to generating markdown reports. Evidence chain: 1. Ingestion points: SKILL.md (Workflow Step 2); 2. Boundary markers: Absent; 3. Capability inventory: Drafting markdown files, no command execution or sensitive file access; 4. Sanitization: Absent.

research-summarizer