The Agent Skills Directory

[SAFE]: The skill documents standard platform features for bot detection and traffic classification. It provides templates for TrendsQuery JSON and HogQL SQL queries that utilize vendor-specific properties like $virt_is_bot and functions such as isLikelyBot. All functions and properties are internal to the PostHog environment.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of untrusted external data (User-Agent strings) to classify traffic.
Ingestion points: The agent is instructed to read properties.$user_agent and properties.$raw_user_agent from the events table in SKILL.md.
Boundary markers: No specific delimiters or "ignore instructions" warnings are included in the query recipes, as the context is analytical data processing rather than prompt construction.
Capability inventory: The skill's capabilities are restricted to data querying and filtering within the PostHog engine. There are no subprocess calls, file-write operations, or network exfiltration patterns detected.
Sanitization: The skill uses built-in HogQL functions (isLikelyBot, getTrafficType) which interpret the User-Agent input as a string for pattern matching, mitigating the risk of instructions within the data being executed.

filtering-bot-traffic