competitor-mapper
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill contains only instructional text and data structures for market research. It does not include executable scripts, shell commands, or network requests.
- [SAFE]: Although the skill instructs the agent to process external data (App Store reviews, Reddit threads, TikTok trends), which is a common surface for indirect prompt injection, it does not possess any high-risk capabilities—such as shell execution, arbitrary file writing to system paths, or network exfiltration—that could be exploited. All output is confined to a specific project directory in JSON format.
- Ingestion points: Reads from
memory/market_insights/containing scraped external content. - Boundary markers: None explicitly defined to separate untrusted data from instructions.
- Capability inventory: Limited to writing a structured JSON file to
memory/ideas/<slug>/competitors.json. - Sanitization: No explicit sanitization or filtering of input data is mentioned in the instructions.
Audit Metadata