The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from the Notion API (such as database metadata or user profiles) which could potentially contain adversarial instructions designed to influence the agent's behavior.
Ingestion points: API responses from api.notion.com processed in the smoke test script (SKILL.md) and pagination helpers (references/code-examples.md).
Boundary markers: The instructions do not define specific delimiters or guardrails to isolate ingested API content from the agent's control logic.
Capability inventory: The skill utilizes Bash (via curl, grep, jq), Read, and Write tools to perform its checks.
Sanitization: The provided property validator focuses on data type integrity but does not include sanitization for potential prompt injection patterns in text strings.
[COMMAND_EXECUTION]: The skill includes a bash-based smoke test script and suggests using grep and git log to audit the local environment for hardcoded credentials. These commands are consistent with the skill's stated purpose of verifying deployment readiness.

notion-prod-checklist