trust-builder
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from multiple sources, creating a potential path for indirect prompt injection. \n
- Ingestion points: The skill uses the browser tool to visit user-supplied base URLs and competitor websites (Phase 2 and Phase 4). It also recursively explores the codebase, reading application routes, API routes, authentication middleware, and business documentation. \n
- Boundary markers: The prompts defined in
references/agent-prompts.mdandreferences/verification-prompts.mddo not include explicit instructions or delimiters to isolate retrieved content or to ignore potential instructions embedded within the target websites or codebase files. \n - Capability inventory: The skill possesses the capability to write files to the local system (
/reports/trust-builder-audit.md) and interact with external web pages through the browser tool. \n - Sanitization: The skill lacks mechanisms to sanitize or filter external content before it is synthesized into the final prioritized opportunity report, making it vulnerable to data poisoning that could alter the agent's recommendations.
Audit Metadata