creating-box-plot-insights

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill generates and executes HogQL (PostHog's SQL dialect) queries at runtime based on user requirements. This involves interpolating user-specified events, properties, and grouping logic into SQL templates to produce distribution data. The skill mitigates risks by instructing the agent to validate the HogQL with posthog:execute-sql and check numeric constraints (e.g., min <= p25 <= median <= p75 <= max) before saving the insight.
  • [INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as it ingests untrusted user requirements and processes data from external tool outputs.
  • Ingestion points: User prompts describing analytics needs and data fetched via posthog:insight-get in SKILL.md.
  • Boundary markers: None defined in the prompt instructions.
  • Capability inventory: Uses posthog:execute-sql, posthog:insight-create, posthog:query-trends, and posthog:insight-query for data operations.
  • Sanitization: The instructions require validating that properties are numeric and verifying the query results before reporting back to the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 07:58 AM
Security Audit — agent-trust-hub — creating-box-plot-insights