The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external integrations (Slack, Gmail, Google Sheets) and the dynamic iPaaS tool catalog, which could contain malicious instructions.
Ingestion points: The skill ingests data via get_ipaas_tools (catalog discovery), call_ipaas_tool (step outputs), and external integration pieces in SKILL.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are mandated for the processed data in the instructions.
Capability inventory: The agent can invoke any tool in the discovered catalog, create connections, and write data to external applications.
Sanitization: The skill relies on host-side validation but lacks explicit agent-side sanitization of the external content before it is interpolated into flow arguments.- [DYNAMIC_EXECUTION]: The skill implements a meta-tool pattern where capabilities are discovered and invoked at runtime rather than being statically defined.
Evidence: The get_ipaas_tools tool fetches the catalog at runtime, and call_ipaas_tool executes these entries using dynamically fetched schemas in SKILL.md.- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: Supports the creation and execution of custom "code steps" within the iPaaS flows.
Evidence: The documentation in SKILL.md describes building flows with "code steps" which allow arbitrary logic execution within the Pipefy iPaaS backend environment.- [DATA_EXPOSURE_AND_EXFILTRATION]: Facilitates data movement between Pipefy and external third-party applications like Slack and Gmail.
Evidence: The primary functionality involves configuring integration pieces and managing OAuth/token connections via create_ipaas_connection in SKILL.md.

pipefy-ipaas