signals-scout-session-replay

Pass

Audited by Gen Agent Trust Hub on Jun 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns or security vulnerabilities were detected. The skill utilizes standard PostHog MCP tools for data analysis and reporting.
  • [PROMPT_INJECTION]: The skill analyzes the risk of indirect prompt injection from user-supplied session content (URLs, element text). It includes a dedicated security section with specific mitigation strategies: 1. Ingestion points: Data enters through execute-sql queries on PostHog event and replay tables in SKILL.md. 2. Boundary markers: A section titled 'Untrusted data' explicitly warns the agent to treat session content as data and never as instructions. 3. Capability inventory: The agent can run SQL queries, emit signals, and manage its own scratchpad memory. 4. Sanitization: The skill mandates the use of sanitized identifiers (slugified/truncated) and instructs the agent to quote and truncate untrusted snippets to prevent interpretation as commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 15, 2026, 12:36 PM
Security Audit — agent-trust-hub — signals-scout-session-replay