The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied content to generate technical diagrams and infographics. While this presents a standard data ingestion surface, the skill implements robust mitigation strategies.
Ingestion points: User requests and source materials provided for transformation into diagrams (SKILL.md).
Boundary markers: The workflow utilizes distinct routing steps (processon-diagram, processon-mindmap, etc.) to isolate different logic flows.
Capability inventory: The skill uses specific MCP tools (generate_chart, generate_diagram, generate_diagram_dsl) which are scoped to visualization rather than general execution.
Sanitization: The instructions include a dedicated prompt optimization step (processon-prompt) and a mandatory review step (processon-review) to validate outputs before final delivery.
[DATA_EXPOSURE]: The skill explicitly defines security boundaries for authentication tokens and sensitive data.
Evidence: Instructions in SKILL.md and references/operations.md strictly prohibit logging, displaying, or persisting credentials like PROCESSON_AUTH_REQUIRED tokens. It also mandates that local attachments can only be uploaded with explicit user authorization.

processon-use