The Agent Skills Directory

[PERSISTENCE]: The skill implements automated reporting by using the gooseworks:automation_upsert tool to create scheduled tasks. These tasks run a dynamically generated prompt on a recurring basis (e.g., every Monday) to perform the analysis without direct user interaction.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data in the form of CSV/Excel exports from platforms like Shopify, Meta, Google Ads, and CRMs.
Ingestion points: Processes order logs, ad performance exports, and CRM lead data.
Boundary markers: Uses specific section headers in reports and structured definitions in definitions.md to scope the agent's focus.
Capability inventory: Access to file_read, file_write, and various MCP tools for platform data retrieval.
Sanitization: The instructions explicitly command the agent to strip PII (customer data) and net out test/staff orders, only keeping totals to avoid data exposure.
[DYNAMIC_EXECUTION]: The skill dynamically assembles and writes an agent_prompt for its automated scheduler. This prompt includes serialized logic and instructions derived from the skill's own reference files, which are copied into the workspace (reports/performance/method/) to facilitate execution in restricted environments without access to the full skills directory.

analyze-performance