brida-reflex-browser-action-risk

Pass

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted browser action descriptions and metadata to classify risk levels, creating a surface where malicious input data could potentially influence the agent's risk assessment.
  • Ingestion points: External inputs such as the action description and consequence summary are used for classification in references/custom-reflex.json and SKILL.md.
  • Boundary markers: The skill includes instructional constraints in references/custom-reflex.json to classify actions "using only the supplied" data, which helps mitigate but does not fully eliminate injection risks.
  • Capability inventory: The skill generates recommendations (automation, review, or block) that are intended to be consumed by downstream browser and computer-use tools.
  • Sanitization: There is no explicit mechanism described for sanitizing or escaping the natural language inputs before processing.
  • [EXTERNAL_DOWNLOADS]: The skill references live configuration and availability data from the official brida-ai vendor resources.
  • Evidence: SKILL.md instructs the agent to read /reflex/llms.txt, /reflex/agent.md, and /reflex/agent.json to determine current service availability and contracts.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 21, 2026, 08:21 PM
Security Audit — agent-trust-hub — brida-reflex-browser-action-risk