The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process data from external sources such as incoming webhooks, Slack messages, and Gmail content. This creates a surface for indirect prompt injection where malicious instructions embedded in that data could influence the agent's behavior.
Ingestion points: Data enters the agent's context through webhooks and integration connectors (Slack, Gmail, Google Sheets) as described in the tools catalog and SKILL.md lifecycle.
Capability inventory: The skill possesses significant capabilities, including writing to external apps, modifying iPaaS data tables, and executing 'code steps' via call_ipaas_tool.
Boundary markers: The instructions do not explicitly mandate the use of delimiters or 'ignore embedded instructions' prompts when handling external data.
Sanitization: There are no specific instructions for the agent to sanitize or validate the content retrieved from external integrations before processing it.
[DYNAMIC_EXECUTION]: The skill enables the creation and execution of integration 'flows' that can include 'code steps' and dynamic routing logic.
Evidence: The call_ipaas_tool meta-tool allows the agent to define and execute these flows on the iPaaS host at runtime. While this is the intended purpose of the skill, it involves the assembly of executable logic based on agent-generated inputs.

pipefy-ipaas