The Agent Skills Directory

[SAFE]: Secure Credential Management. The skill provides an .env.example template and explicitly instructs users to store API keys in environment variables rather than hardcoding them, which is the recommended practice for sensitive secrets.
[SAFE]: Resource Scoping and Least Privilege. The documentation recommends using read-only API scopes (e.g., flows:read, campaigns:read) for analyst tasks, minimizing the potential impact of any credential exposure.
[SAFE]: Path Validation. The included Python scripts (analyze.py, klaviyo_client.py) implement a _safe_output_path function that resolves real paths and ensures output files are written only within the current working directory, effectively preventing directory traversal attacks.
[SAFE]: Rate Limiting and Performance Safety. The analysis scripts include iteration caps (MAX_FLOW_REPORTS, MAX_CAMPAIGN_REPORTS) and utilize the SDK's built-in retry logic. This prevents the skill from exhausting the API budget or causing accidental denial-of-service on large accounts.
[SAFE]: Indirect Prompt Injection Analysis. The skill possesses an ingestion surface for external data via the Klaviyo API (e.g., flow and campaign names). However, the scripts treat this data strictly as strings for aggregation and reporting, and the logic does not involve unsafe interpolation into prompts or execution of the ingested data.
Ingestion points: Data enters the environment through klaviyo_client.py calling the Klaviyo API (a.klaviyo.com).
Boundary markers: Absent; the skill relies on standard data processing patterns.
Capability inventory: Execution of local analysis scripts, network access to official Klaviyo API endpoints, and local file writing for reports.
Sanitization: Data is parsed as JSON and flattened for statistical reporting; it is not evaluated as code or command input.

klaviyo-analyst